1 systemd System and Service Manager
5 * journalctl gained the new options -t/--identifier= to match
6 on the syslog identifier (aka "tag"), as well as --utc to
7 show log timestamps in the UTC timezone. journalctl now also
8 accepts -n/--lines=all to disable line capping in a pager.
10 * journalctl gained a new switch, --flush, that synchronously
11 flushes logs from /run/log/journal to /var/log/journal if
12 persistent storage is enabled. systemd-journal-flush.service
13 now waits until the operation is complete.
15 * Services can notify the manager before they start a reload
16 (by sending RELOADING=1) or shutdown (by sending
17 STOPPING=1). This allows the manager to track and show the
18 internal state of daemons and closes a race condition when
19 the process is still running but has closed its D-Bus
22 * Services with Type=oneshot do not have to have any ExecStart
25 * User units are now loaded also from
26 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
27 /run/systemd/user directory that was already previously
28 supported, but is under the control of the user.
30 * Job timeouts (i.e. time-outs on the time a job that is
31 queued stays in the run queue) can now optionally result in
32 immediate reboot or power-off actions (JobTimeoutAction= and
33 JobTimeoutRebootArgument=). This is useful on ".target"
34 units, to limit the maximum time a target remains
35 undispatched in the run queue, and to trigger an emergency
36 operation in such a case. This is now used by default to
37 turn off the system if boot-up (as defined by everything in
38 basic.target) hangs and does not complete for at least
39 15min. Also, if power-off or reboot hang for at least 30min
40 an immediate power-off/reboot operation is triggered. This
41 functionality is particularly useful to increase reliability
42 on embedded devices, but also on laptops which might
43 accidentally get powered on when carried in a backpack and
44 whose boot stays stuck in a hard disk encryption passphrase
47 * systemd-logind can be configured to also handle lid switch
48 events even when the machine is docked or multiple displays
49 are attached (HandleLidSwitchDocked= option).
51 * A helper binary and a service have been added which can be
52 used to resume from hibernation in the initramfs. A
53 generator will parse the resume= option on the kernel
54 command-line to trigger resume.
56 * A user console daemon systemd-consoled has been
57 added. Currently, it is a preview, and will so far open a
58 single terminal on each session of the user marked as
59 Desktop=systemd-console.
61 * Route metrics can be specified for DHCP routes added by
64 * The SELinux context of socket-activated services can be set
65 from the information provided by the networking stack
66 (SELinuxContextFromNet= option).
68 * Userspace firmware loading support has been removed and
69 the minimum supported kernel version is thus bumped to 3.7.
71 * Timeout for udev workers has been increased from 1 to 3
72 minutes, but a warning will be printed after 1 minute to
73 help diagnose kernel modules that take a long time to load.
75 * Udev rules can now remove tags on devices with TAG-="foobar".
77 * systemd's readahead implementation has been removed. In many
78 circumstances it didn't give expected benefits even for
79 rotational disk drives and was becoming less relevant in the
80 age of SSDs. As none of the developers has been using
81 rotating media anymore, and nobody stepped up to actively
82 maintain this component of systemd it has now been removed.
84 * Swap units can use Options= to specify discard options.
85 Discard options specified for swaps in /etc/fstab are now
88 * Docker containers are now detected as a separate type of
91 * The Password Agent protocol gained support for queries where
92 the user input is shown, useful e.g. for user names.
93 systemd-ask-password gained a new --echo option to turn that
96 * The default sysctl.d/ snippets will now set:
98 net.core.default_qdisc = fq_codel
100 This selects Fair Queuing Controlled Delay as the default
101 queuing discipline for network interfaces. fq_codel helps
102 fight the network bufferbloat problem. It is believed to be
103 a good default with no tuning required for most workloads.
104 Downstream distributions may override this choice. On 10Gbit
105 servers that do not do forwarding, "fq" may perform better.
106 Systems without a good clocksource should use "pfifo_fast".
108 * If kdbus is enabled during build a new option BusPolicy= is
109 available for service units, that allows locking all service
110 processes into a stricter bus policy, in order to limit
111 access to various bus services, or even hide most of them
112 from the service's view entirely.
114 * networkctl will now show the .network and .link file
115 networkd has applied to a specific interface.
117 * sd-login gained a new API call sd_session_get_desktop() to
118 query which desktop environment has been selected for a
121 * UNIX utmp support is now compile-time optional to support
124 * systemctl gained two new commands "add-wants" and
125 "add-requires" for pulling in units from specific targets
128 * If the word "rescue" is specified on the kernel command line
129 the system will now boot into rescue mode (aka
130 rescue.target), which was previously available only by
131 specifying "1" or "systemd.unit=rescue.target" on the kernel
132 command line. This new kernel command line option nicely
133 mirrors the already existing "emergency" kernel command line
136 * New kernel command line options mount.usr=, mount.usrflags=,
137 mount.usrfstype= have been added that match root=, rootflags=,
138 rootfstype= but allow mounting a specific file system to
141 * The $NOTIFY_SOCKET is now also passed to control processes of
142 services, not only the main process.
144 * This version reenables support for fsck's -l switch. This
145 means at least version v2.25 of util-linux is required for
146 operation, otherwise dead-locks on device nodes may
147 occur. Again: you need to update util-linux to at least
148 v2.25 when updating systemd to v217.
150 * The "multi-seat-x" tool has been removed from systemd, as
151 its functionality has been integrated into X servers 1.16,
152 and the tool is hence redundant. It is recommended to update
153 display managers invoking this tool to simply invoke X
154 directly from now on, again.
156 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
157 message flag has been added for all of systemd's PolicyKit
158 authenticated method calls has been added. In particular
159 this now allows optional interactive authorization via
160 PolicyKit for many of PID1's privileged operations such as
161 unit file enabling and disabling.
163 * "udevadm hwdb --update" learnt a new switch "--usr" for
164 placing the rebuilt hardware database in /usr instead of
165 /etc. When used only hardware database entries stored in
166 /usr will be used, and any user database entries in /etc are
167 ignored. This functionality is useful for vendors to ship a
168 pre-built database on systems where local configuration is
169 unnecessary or unlikely.
171 * Calendar time specifications in .timer units now also
172 understand the strings "semi-annually", "quarterly" and
173 "minutely" as shortcuts (in addition to the preexisting
174 "anually", "hourly", ...).
176 * systemd-tmpfiles will now correctly create files in /dev
177 at boot which are marked for creation only at boot. It is
178 recommended to always create static device nodes with 'c!'
179 and 'b!', so that they are created only at boot and not
180 overwritten at runtime.
182 * When the watchdog logic is used for a service (WatchdogSec=)
183 and the watchdog timeout is hit the service will now be
184 terminated with SIGABRT (instead of just SIGTERM), in order
185 to make sure a proper coredump and backtrace is
186 generated. This ensures that hanging services will result in
187 similar coredump/backtrace behaviour as services that hit a
190 Contributions from: Andreas Henriksson, Andrei Borzenkov,
191 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
192 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
193 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
194 Herrmann, David Sommerseth, David Strauss, Emil Renner
195 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
196 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
197 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
198 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
199 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
200 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
201 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
202 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
203 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
204 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
205 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
206 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
207 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
208 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
209 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
210 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
213 -- Berlin, 2014-10-28
217 * timedated no longer reads NTP implementation unit names from
218 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
219 implementations should add a
221 Conflicts=systemd-timesyncd.service
223 to their unit files to take over and replace systemd's NTP
224 default functionality.
226 * systemd-sysusers gained a new line type "r" for configuring
227 which UID/GID ranges to allocate system users/groups
228 from. Lines of type "u" may now add an additional column
229 that specifies the home directory for the system user to be
230 created. Also, systemd-sysusers may now optionally read user
231 information from STDIN instead of a file. This is useful for
232 invoking it from RPM preinst scriptlets that need to create
233 users before the first RPM file is installed since these
234 files might need to be owned by them. A new
235 %sysusers_create_inline RPM macro has been introduced to do
236 just that. systemd-sysusers now updates the shadow files as
237 well as the user/group databases, which should enhance
238 compatibility with certain tools like grpck.
240 * A number of bus APIs of PID 1 now optionally consult
241 PolicyKit to permit access for otherwise unprivileged
242 clients under certain conditions. Note that this currently
243 doesn't support interactive authentication yet, but this is
244 expected to be added eventually, too.
246 * /etc/machine-info now has new fields for configuring the
247 deployment environment of the machine, as well as the
248 location of the machine. hostnamectl has been updated with
249 new command to update these fields.
251 * systemd-timesyncd has been updated to automatically acquire
252 NTP server information from systemd-networkd, which might
253 have been discovered via DHCP.
255 * systemd-resolved now includes a caching DNS stub resolver
256 and a complete LLMNR name resolution implementation. A new
257 NSS module "nss-resolve" has been added which can be used
258 instead of glibc's own "nss-dns" to resolve hostnames via
259 systemd-resolved. Hostnames, addresses and arbitrary RRs may
260 be resolved via systemd-resolved D-Bus APIs. In contrast to
261 the glibc internal resolver systemd-resolved is aware of
262 multi-homed system, and keeps DNS server and caches separate
263 and per-interface. Queries are sent simultaneously on all
264 interfaces that have DNS servers configured, in order to
265 properly handle VPNs and local LANs which might resolve
266 separate sets of domain names. systemd-resolved may acquire
267 DNS server information from systemd-networkd automatically,
268 which in turn might have discovered them via DHCP. A tool
269 "systemd-resolve-host" has been added that may be used to
270 query the DNS logic in resolved. systemd-resolved implements
271 IDNA and automatically uses IDNA or UTF-8 encoding depending
272 on whether classic DNS or LLMNR is used as transport. In the
273 next releases we intend to add a DNSSEC and mDNS/DNS-SD
274 implementation to systemd-resolved.
276 * A new NSS module nss-mymachines has been added, that
277 automatically resolves the names of all local registered
278 containers to their respective IP addresses.
280 * A new client tool "networkctl" for systemd-networkd has been
281 added. It currently is entirely passive and will query
282 networking configuration from udev, rtnetlink and networkd,
283 and present it to the user in a very friendly
284 way. Eventually, we hope to extend it to become a full
285 control utility for networkd.
287 * .socket units gained a new DeferAcceptSec= setting that
288 controls the kernels' TCP_DEFER_ACCEPT sockopt for
289 TCP. Similar, support for controlling TCP keep-alive
290 settings has been added (KeepAliveTimeSec=,
291 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
292 turning off Nagle's algorithm on TCP has been added
295 * logind learned a new session type "web", for use in projects
296 like Cockpit which register web clients as PAM sessions.
298 * timer units with at least one OnCalendar= setting will now
299 be started only after timer-sync.target has been
300 reached. This way they will not elapse before the system
301 clock has been corrected by a local NTP client or
302 similar. This is particular useful on RTC-less embedded
303 machines, that come up with an invalid system clock.
305 * systemd-nspawn's --network-veth= switch should now result in
306 stable MAC addresses for both the outer and the inner side
309 * systemd-nspawn gained a new --volatile= switch for running
310 container instances with /etc or /var unpopulated.
312 * The kdbus client code has been updated to use the new Linux
313 3.17 memfd subsystem instead of the old kdbus-specific one.
315 * systemd-networkd's DHCP client and server now support
316 FORCERENEW. There are also new configuration options to
317 configure the vendor client identifier and broadcast mode
320 * systemd will no longer inform the kernel about the current
321 timezone, as this is necessarily incorrect and racy as the
322 kernel has no understanding of DST and similar
323 concepts. This hence means FAT timestamps will be always
324 considered UTC, similar to what Android is already
325 doing. Also, when the RTC is configured to the local time
326 (rather than UTC) systemd will never synchronize back to it,
327 as this might confuse Windows at a later boot.
329 * systemd-analyze gained a new command "verify" for offline
330 validation of unit files.
332 * systemd-networkd gained support for a couple of additional
333 settings for bonding networking setups. Also, the metric for
334 statically configured routes may now be configured. For
335 network interfaces where this is appropriate the peer IP
336 address may now be configured.
338 * systemd-networkd's DHCP client will no longer request
339 broadcasting by default, as this tripped up some networks.
340 For hardware where broadcast is required the feature should
341 be switched back on using RequestBroadcast=yes.
343 * systemd-networkd will now set up IPv4LL addresses (when
344 enabled) even if DHCP is configured successfully.
346 * udev will now default to respect network device names given
347 by the kernel when the kernel indicates that these are
348 predictable. This behavior can be tweaked by changing
349 NamePolicy= in the relevant .link file.
351 * A new library systemd-terminal has been added that
352 implements full TTY stream parsing and rendering. This
353 library is supposed to be used later on for implementing a
354 full userspace VT subsystem, replacing the current kernel
357 * A new tool systemd-journal-upload has been added to push
358 journal data to a remote system running
359 systemd-journal-remote.
361 * journald will no longer forward all local data to another
362 running syslog daemon. This change has been made because
363 rsyslog (which appears to be the most commonly used syslog
364 implementation these days) no longer makes use of this, and
365 instead pulls the data out of the journal on its own. Since
366 forwarding the messages to a non-existent syslog server is
367 more expensive than we assumed we have now turned this
368 off. If you run a syslog server that is not a recent rsyslog
369 version, you have to turn this option on again
370 (ForwardToSyslog= in journald.conf).
372 * journald now optionally supports the LZ4 compressor for
373 larger journal fields. This compressor should perform much
374 better than XZ which was the previous default.
376 * machinectl now shows the IP addresses of local containers,
377 if it knows them, plus the interface name of the container.
379 * A new tool "systemd-escape" has been added that makes it
380 easy to escape strings to build unit names and similar.
382 * sd_notify() messages may now include a new ERRNO= field
383 which is parsed and collected by systemd and shown among the
384 "systemctl status" output for a service.
386 * A new component "systemd-firstboot" has been added that
387 queries the most basic systemd information (timezone,
388 hostname, root password) interactively on first
389 boot. Alternatively it may also be used to provision these
390 things offline on OS images installed into directories.
392 * The default sysctl.d/ snippets will now set
394 net.ipv4.conf.default.promote_secondaries=1
396 This has the benefit of no flushing secondary IP addresses
397 when primary addresses are removed.
399 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
400 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
401 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
402 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
403 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
404 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
405 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
406 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
407 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
408 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
409 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
410 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
411 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
412 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
413 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
415 -- Berlin, 2014-08-19
419 * A new tool systemd-sysusers has been added. This tool
420 creates system users and groups in /etc/passwd and
421 /etc/group, based on static declarative system user/group
422 definitions in /usr/lib/sysusers.d/. This is useful to
423 enable factory resets and volatile systems that boot up with
424 an empty /etc directory, and thus need system users and
425 groups created during early boot. systemd now also ships
426 with two default sysusers.d/ files for the most basic
427 users and groups systemd and the core operating system
430 * A new tmpfiles snippet has been added that rebuilds the
431 essential files in /etc on boot, should they be missing.
433 * A directive for ensuring automatic clean-up of
434 /var/cache/man/ has been removed from the default
435 configuration. This line should now be shipped by the man
436 implementation. The necessary change has been made to the
437 man-db implementation. Note that you need to update your man
438 implementation to one that ships this line, otherwise no
439 automatic clean-up of /var/cache/man will take place.
441 * A new condition ConditionNeedsUpdate= has been added that
442 may conditionalize services to only run when /etc or /var
443 are "older" than the vendor operating system resources in
444 /usr. This is useful for reconstructing or updating /etc
445 after an offline update of /usr or a factory reset, on the
446 next reboot. Services that want to run once after such an
447 update or reset should use this condition and order
448 themselves before the new systemd-update-done.service, which
449 will mark the two directories as fully updated. A number of
450 service files have been added making use of this, to rebuild
451 the udev hardware database, the journald message catalog and
452 dynamic loader cache (ldconfig). The systemd-sysusers tool
453 described above also makes use of this now. With this in
454 place it is now possible to start up a minimal operating
455 system with /etc empty cleanly. For more information on the
456 concepts involved see this recent blog story:
458 http://0pointer.de/blog/projects/stateless.html
460 * A new system group "input" has been introduced, and all
461 input device nodes get this group assigned. This is useful
462 for system-level software to get access to input devices. It
463 complements what is already done for "audio" and "video".
465 * systemd-networkd learnt minimal DHCPv4 server support in
466 addition to the existing DHCPv4 client support. It also
467 learnt DHCPv6 client and IPv6 Router Solicitation client
468 support. The DHCPv4 client gained support for static routes
469 passed in from the server. Note that the [DHCPv4] section
470 known in older systemd-networkd versions has been renamed to
471 [DHCP] and is now also used by the DHCPv6 client. Existing
472 .network files using settings of this section should be
473 updated, though compatibility is maintained. Optionally, the
474 client hostname may now be sent to the DHCP server.
476 * networkd gained support for vxlan virtual networks as well
477 as tun/tap and dummy devices.
479 * networkd gained support for automatic allocation of address
480 ranges for interfaces from a system-wide pool of
481 addresses. This is useful for dynamically managing a large
482 number of interfaces with a single network configuration
483 file. In particular this is useful to easily assign
484 appropriate IP addresses to the veth links of a large number
487 * RPM macros for processing sysusers, sysctl and binfmt
488 drop-in snippets at package installation time have been
491 * The /etc/os-release file should now be placed in
492 /usr/lib/os-release. The old location is automatically
493 created as symlink. /usr/lib is the more appropriate
494 location of this file, since it shall actually describe the
495 vendor operating system shipped in /usr, and not the
496 configuration stored in /etc.
498 * .mount units gained a new boolean SloppyOptions= setting
499 that maps to mount(8)'s -s option which enables permissive
500 parsing of unknown mount options.
502 * tmpfiles learnt a new "L+" directive which creates a symlink
503 but (unlike "L") deletes a pre-existing file first, should
504 it already exist and not already be the correct
505 symlink. Similar, "b+", "c+" and "p+" directives have been
506 added as well, which create block and character devices, as
507 well as fifos in the filesystem, possibly removing any
508 pre-existing files of different types.
510 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
511 'argument' field (which so far specified the source to
512 symlink/copy the files from) is now optional. If omitted the
513 same file os copied from /usr/share/factory/ suffixed by the
514 full destination path. This is useful for populating /etc
515 with essential files, by copying them from vendor defaults
516 shipped in /usr/share/factory/etc.
518 * A new command "systemctl preset-all" has been added that
519 applies the service preset settings to all installed unit
520 files. A new switch --preset-mode= has been added that
521 controls whether only enable or only disable operations
524 * A new command "systemctl is-system-running" has been added
525 that allows checking the overall state of the system, for
526 example whether it is fully up and running.
528 * When the system boots up with an empty /etc, the equivalent
529 to "systemctl preset-all" is executed during early boot, to
530 make sure all default services are enabled after a factory
533 * systemd now contains a minimal preset file that enables the
534 most basic services systemd ships by default.
536 * Unit files' [Install] section gained a new DefaultInstance=
537 field for defining the default instance to create if a
538 template unit is enabled with no instance specified.
540 * A new passive target cryptsetup-pre.target has been added
541 that may be used by services that need to make they run and
542 finish before the first LUKS cryptographic device is set up.
544 * The /dev/loop-control and /dev/btrfs-control device nodes
545 are now owned by the "disk" group by default, opening up
546 access to this group.
548 * systemd-coredump will now automatically generate a
549 stack trace of all core dumps taking place on the system,
550 based on elfutils' libdw library. This stack trace is logged
553 * systemd-coredump may now optionally store coredumps directly
554 on disk (in /var/lib/systemd/coredump, possibly compressed),
555 instead of storing them unconditionally in the journal. This
556 mode is the new default. A new configuration file
557 /etc/systemd/coredump.conf has been added to configure this
558 and other parameters of systemd-coredump.
560 * coredumpctl gained a new "info" verb to show details about a
561 specific coredump. A new switch "-1" has also been added
562 that makes sure to only show information about the most
563 recent entry instead of all entries. Also, as the tool is
564 generally useful now the "systemd-" prefix of the binary
565 name has been removed. Distributions that want to maintain
566 compatibility with the old name should add a symlink from
567 the old name to the new name.
569 * journald's SplitMode= now defaults to "uid". This makes sure
570 that unprivileged users can access their own coredumps with
571 coredumpctl without restrictions.
573 * New kernel command line options "systemd.wants=" (for
574 pulling an additional unit during boot), "systemd.mask="
575 (for masking a specific unit for the boot), and
576 "systemd.debug-shell" (for enabling the debug shell on tty9)
577 have been added. This is implemented in the new generator
578 "systemd-debug-generator".
580 * systemd-nspawn will now by default filter a couple of
581 syscalls for containers, among them those required for
582 kernel module loading, direct x86 IO port access, swap
583 management, and kexec. Most importantly though
584 open_by_handle_at() is now prohibited for containers,
585 closing a hole similar to a recently discussed vulnerability
586 in docker regarding access to files on file hierarchies the
587 container should normally not have access to. Note that for
588 nspawn we generally make no security claims anyway (and
589 this is explicitly documented in the man page), so this is
590 just a fix for one of the most obvious problems.
592 * A new man page file-hierarchy(7) has been added that
593 contains a minimized, modernized version of the file system
594 layout systemd expects, similar in style to the FHS
595 specification or hier(5). A new tool systemd-path(1) has
596 been added to query many of these paths for the local
599 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
600 longer done. Since the directory now has a per-user size
601 limit, and is cleaned on logout this appears unnecessary,
602 in particular since this now brings the lifecycle of this
603 directory closer in line with how IPC objects are handled.
605 * systemd.pc now exports a number of additional directories,
606 including $libdir (which is useful to identify the library
607 path for the primary architecture of the system), and a
608 couple of drop-in directories.
610 * udev's predictable network interface names now use the dev_port
611 sysfs attribute, introduced in linux 3.15 instead of dev_id to
612 distinguish between ports of the same PCI function. dev_id should
613 only be used for ports using the same HW address, hence the need
616 * machined has been updated to export the OS version of a
617 container (read from /etc/os-release and
618 /usr/lib/os-release) on the bus. This is now shown in
619 "machinectl status" for a machine.
621 * A new service setting RestartForceExitStatus= has been
622 added. If configured to a set of exit signals or process
623 return values, the service will be restarted when the main
624 daemon process exits with any of them, regardless of the
627 * systemctl's -H switch for connecting to remote systemd
628 machines has been extended so that it may be used to
629 directly connect to a specific container on the
630 host. "systemctl -H root@foobar:waldi" will now connect as
631 user "root" to host "foobar", and then proceed directly to
632 the container named "waldi". Note that currently you have to
633 authenticate as user "root" for this to work, as entering
634 containers is a privileged operation.
636 Contributions from: Andreas Henriksson, Benjamin Steinwender,
637 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
638 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
639 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
640 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
641 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
642 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
643 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
644 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
645 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
646 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
647 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
649 -- Berlin, 2014-07-03
653 * As an experimental feature, udev now tries to lock the
654 disk device node (flock(LOCK_SH|LOCK_NB)) while it
655 executes events for the disk or any of its partitions.
656 Applications like partitioning programs can lock the
657 disk device node (flock(LOCK_EX)) and claim temporary
658 device ownership that way; udev will entirely skip all event
659 handling for this disk and its partitions. If the disk
660 was opened for writing, the close will trigger a partition
661 table rescan in udev's "watch" facility, and if needed
662 synthesize "change" events for the disk and all its partitions.
663 This is now unconditionally enabled, and if it turns out to
664 cause major problems, we might turn it on only for specific
665 devices, or might need to disable it entirely. Device Mapper
666 devices are excluded from this logic.
668 * We temporarily dropped the "-l" switch for fsck invocations,
669 since they collide with the flock() logic above. util-linux
670 upstream has been changed already to avoid this conflict,
671 and we will readd "-l" as soon as util-linux with this
672 change has been released.
674 * The dependency on libattr has been removed. Since a long
675 time, the extended attribute calls have moved to glibc, and
676 libattr is thus unnecessary.
678 * Virtualization detection works without priviliges now. This
679 means the systemd-detect-virt binary no longer requires
680 CAP_SYS_PTRACE file capabilities, and our daemons can run
681 with fewer privileges.
683 * systemd-networkd now runs under its own "systemd-network"
684 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
685 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
686 loses the ability to write to files owned by root this way.
688 * Similar, systemd-resolved now runs under its own
689 "systemd-resolve" user with no capabilities remaining.
691 * Similar, systemd-bus-proxyd now runs under its own
692 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
694 * systemd-networkd gained support for setting up "veth"
695 virtual ethernet devices for container connectivity, as well
696 as GRE and VTI tunnels.
698 * systemd-networkd will no longer automatically attempt to
699 manually load kernel modules necessary for certain tunnel
700 transports. Instead, it is assumed the kernel loads them
701 automatically when required. This only works correctly on
702 very new kernels. On older kernels, please consider adding
703 the kernel modules to /etc/modules-load.d/ as a work-around.
705 * The resolv.conf file systemd-resolved generates has been
706 moved to /run/systemd/resolve/. If you have a symlink from
707 /etc/resolv.conf, it might be necessary to correct it.
709 * Two new service settings, ProtectHome= and ProtectSystem=,
710 have been added. When enabled, they will make the user data
711 (such as /home) inaccessible or read-only and the system
712 (such as /usr) read-only, for specific services. This allows
713 very light-weight per-service sandboxing to avoid
714 modifications of user data or system files from
715 services. These two new switches have been enabled for all
716 of systemd's long-running services, where appropriate.
718 * Socket units gained new SocketUser= and SocketGroup=
719 settings to set the owner user and group of AF_UNIX sockets
720 and FIFOs in the file system.
722 * Socket units gained a new RemoveOnStop= setting. If enabled,
723 all FIFOS and sockets in the file system will be removed
724 when the specific socket unit is stopped.
726 * Socket units gained a new Symlinks= setting. It takes a list
727 of symlinks to create to file system sockets or FIFOs
728 created by the specific Unix sockets. This is useful to
729 manage symlinks to socket nodes with the same life-cycle as
732 * The /dev/log socket and /dev/initctl FIFO have been moved to
733 /run, and have been replaced by symlinks. This allows
734 connecting to these facilities even if PrivateDevices=yes is
735 used for a service (which makes /dev/log itself unavailable,
736 but /run is left). This also has the benefit of ensuring
737 that /dev only contains device nodes, directories and
738 symlinks, and nothing else.
740 * sd-daemon gained two new calls sd_pid_notify() and
741 sd_pid_notifyf(). They are similar to sd_notify() and
742 sd_notifyf(), but allow overriding of the source PID of
743 notification messages if permissions permit this. This is
744 useful to send notify messages on behalf of a different
745 process (for example, the parent process). The
746 systemd-notify tool has been updated to make use of this
747 when sending messages (so that notification messages now
748 originate from the shell script invoking systemd-notify and
749 not the systemd-notify process itself. This should minimize
750 a race where systemd fails to associate notification
751 messages to services when the originating process already
754 * A new "on-abnormal" setting for Restart= has been added. If
755 set, it will result in automatic restarts on all "abnormal"
756 reasons for a process to exit, which includes unclean
757 signals, core dumps, timeouts and watchdog timeouts, but
758 does not include clean and unclean exit codes or clean
759 signals. Restart=on-abnormal is an alternative for
760 Restart=on-failure for services that shall be able to
761 terminate and avoid restarts on certain errors, by
762 indicating so with an unclean exit code. Restart=on-failure
763 or Restart=on-abnormal is now the recommended setting for
764 all long-running services.
766 * If the InaccessibleDirectories= service setting points to a
767 mount point (or if there are any submounts contained within
768 it), it is now attempted to completely unmount it, to make
769 the file systems truly unavailable for the respective
772 * The ReadOnlyDirectories= service setting and
773 systemd-nspawn's --read-only parameter are now recursively
774 applied to all submounts, too.
776 * Mount units may now be created transiently via the bus APIs.
778 * The support for SysV and LSB init scripts has been removed
779 from the systemd daemon itself. Instead, it is now
780 implemented as a generator that creates native systemd units
781 from these scripts when needed. This enables us to remove a
782 substantial amount of legacy code from PID 1, following the
783 fact that many distributions only ship a very small number
784 of LSB/SysV init scripts nowadays.
786 * Priviliged Xen (dom0) domains are not considered
787 virtualization anymore by the virtualization detection
788 logic. After all, they generally have unrestricted access to
789 the hardware and usually are used to manage the unprivileged
792 * systemd-tmpfiles gained a new "C" line type, for copying
793 files or entire directories.
795 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
796 lines. So far, they have been non-globbing versions of the
797 latter, and have thus been redundant. In future, it is
798 recommended to only use "z". "m" has hence been removed
799 from the documentation, even though it stays supported.
801 * A tmpfiles snippet to recreate the most basic structure in
802 /var has been added. This is enough to create the /var/run →
803 /run symlink and create a couple of structural
804 directories. This allows systems to boot up with an empty or
805 volatile /var. Of course, while with this change, the core OS
806 now is capable with dealing with a volatile /var, not all
807 user services are ready for it. However, we hope that sooner
808 or later, many service daemons will be changed upstream so
809 that they are able to automatically create their necessary
810 directories in /var at boot, should they be missing. This is
811 the first step to allow state-less systems that only require
812 the vendor image for /usr to boot.
814 * systemd-nspawn has gained a new --tmpfs= switch to mount an
815 empty tmpfs instance to a specific directory. This is
816 particularly useful for making use of the automatic
817 reconstruction of /var (see above), by passing --tmpfs=/var.
819 * Access modes specified in tmpfiles snippets may now be
820 prefixed with "~", which indicates that they shall be masked
821 by whether the existing file or directory is currently
822 writable, readable or executable at all. Also, if specified,
823 the sgid/suid/sticky bits will be masked for all
826 * A new passive target unit "network-pre.target" has been
827 added which is useful for services that shall run before any
828 network is configured, for example firewall scripts.
830 * The "floppy" group that previously owned the /dev/fd*
831 devices is no longer used. The "disk" group is now used
832 instead. Distributions should probably deprecate usage of
835 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
836 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
837 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
838 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
839 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
840 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
843 -- Berlin, 2014-06-11
847 * A new "systemd-timesyncd" daemon has been added for
848 synchronizing the system clock across the network. It
849 implements an SNTP client. In contrast to NTP
850 implementations such as chrony or the NTP reference server,
851 this only implements a client side, and does not bother with
852 the full NTP complexity, focusing only on querying time from
853 one remote server and synchronizing the local clock to
854 it. Unless you intend to serve NTP to networked clients or
855 want to connect to local hardware clocks, this simple NTP
856 client should be more than appropriate for most
857 installations. The daemon runs with minimal privileges, and
858 has been hooked up with networkd to only operate when
859 network connectivity is available. The daemon saves the
860 current clock to disk every time a new NTP sync has been
861 acquired, and uses this to possibly correct the system clock
862 early at bootup, in order to accommodate for systems that
863 lack an RTC such as the Raspberry Pi and embedded devices,
864 and to make sure that time monotonically progresses on these
865 systems, even if it is not always correct. To make use of
866 this daemon, a new system user and group "systemd-timesync"
867 needs to be created on installation of systemd.
869 * The queue "seqnum" interface of libudev has been disabled, as
870 it was generally incompatible with device namespacing as
871 sequence numbers of devices go "missing" if the devices are
872 part of a different namespace.
874 * "systemctl list-timers" and "systemctl list-sockets" gained
875 a --recursive switch for showing units of these types also
876 for all local containers, similar in style to the already
877 supported --recursive switch for "systemctl list-units".
879 * A new RebootArgument= setting has been added for service
880 units, which may be used to specify a kernel reboot argument
881 to use when triggering reboots with StartLimitAction=.
883 * A new FailureAction= setting has been added for service
884 units which may be used to specify an operation to trigger
885 when a service fails. This works similarly to
886 StartLimitAction=, but unlike it, controls what is done
887 immediately rather than only after several attempts to
888 restart the service in question.
890 * hostnamed got updated to also expose the kernel name,
891 release, and version on the bus. This is useful for
892 executing commands like hostnamectl with the -H switch.
893 systemd-analyze makes use of this to properly display
894 details when running non-locally.
896 * The bootchart tool can now show cgroup information in the
899 * The CFS CPU quota cgroup attribute is now exposed for
900 services. The new CPUQuota= switch has been added for this
901 which takes a percentage value. Setting this will have the
902 result that a service may never get more CPU time than the
903 specified percentage, even if the machine is otherwise idle.
905 * systemd-networkd learned IPIP and SIT tunnel support.
907 * LSB init scripts exposing a dependency on $network will now
908 get a dependency on network-online.target rather than simply
909 network.target. This should bring LSB handling closer to
910 what it was on SysV systems.
912 * A new fsck.repair= kernel option has been added to control
913 how fsck shall deal with unclean file systems at boot.
915 * The (.ini) configuration file parser will now silently
916 ignore sections whose name begins with "X-". This may be
917 used to maintain application-specific extension sections in unit
920 * machined gained a new API to query the IP addresses of
921 registered containers. "machinectl status" has been updated
922 to show these addresses in its output.
924 * A new call sd_uid_get_display() has been added to the
925 sd-login APIs for querying the "primary" session of a
926 user. The "primary" session of the user is elected from the
927 user's sessions and generally a graphical session is
928 preferred over a text one.
930 * A minimal systemd-resolved daemon has been added. It
931 currently simply acts as a companion to systemd-networkd and
932 manages resolv.conf based on per-interface DNS
933 configuration, possibly supplied via DHCP. In the long run
934 we hope to extend this into a local DNSSEC enabled DNS and
937 * The systemd-networkd-wait-online tool is now enabled by
938 default. It will delay network-online.target until a network
939 connection has been configured. The tool primarily integrates
940 with networkd, but will also make a best effort to make sense
941 of network configuration performed in some other way.
943 * Two new service options StartupCPUShares= and
944 StartupBlockIOWeight= have been added that work similarly to
945 CPUShares= and BlockIOWeight= however only apply during
946 system startup. This is useful to prioritize certain services
947 differently during bootup than during normal runtime.
949 * hostnamed has been changed to prefer the statically
950 configured hostname in /etc/hostname (unless set to
951 'localhost' or empty) over any dynamic one supplied by
952 dhcp. With this change, the rules for picking the hostname
953 match more closely the rules of other configuration settings
954 where the local administrator's configuration in /etc always
955 overrides any other settings.
957 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
958 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
959 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
960 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
961 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
962 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
963 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
964 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
965 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
966 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
967 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
968 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
969 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
970 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
971 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
972 Lindskog, WaLyong Cho, Will Woods, Zbigniew
975 -- Beijing, 2014-05-28
979 * When restoring the screen brightness at boot, stay away from
980 the darkest setting or from the lowest 5% of the available
981 range, depending on which is the larger value of both. This
982 should effectively protect the user from rebooting into a
983 black screen, should the brightness have been set to minimum
986 * sd-login gained a new sd_machine_get_class() call to
987 determine the class ("vm" or "container") of a machine
988 registered with machined.
990 * sd-login gained new calls
991 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
992 to query the identity of the peer of a local AF_UNIX
993 connection. They operate similarly to their sd_pid_get_xyz()
996 * PID 1 will now maintain a system-wide system state engine
997 with the states "starting", "running", "degraded",
998 "maintenance", "stopping". These states are bound to system
999 startup, normal runtime, runtime with at least one failed
1000 service, rescue/emergency mode and system shutdown. This
1001 state is shown in the "systemctl status" output when no unit
1002 name is passed. It is useful to determine system state, in
1003 particularly when doing so for many systems or containers at
1006 * A new command "list-machines" has been added to "systemctl"
1007 that lists all local OS containers and shows their system
1008 state (see above), if systemd runs inside of them.
1010 * systemctl gained a new "-r" switch to recursively enumerate
1011 units on all local containers, when used with the
1012 "list-unit" command (which is the default one that is
1013 executed when no parameters are specified).
1015 * The GPT automatic partition discovery logic will now honour
1016 two GPT partition flags: one may be set on a partition to
1017 cause it to be mounted read-only, and the other may be set
1018 on a partition to ignore it during automatic discovery.
1020 * Two new GPT type UUIDs have been added for automatic root
1021 partition discovery, for 32-bit and 64-bit ARM. This is not
1022 particularly useful for discovering the root directory on
1023 these architectures during bare-metal boots (since UEFI is
1024 not common there), but still very useful to allow booting of
1025 ARM disk images in nspawn with the -i option.
1027 * MAC addresses of interfaces created with nspawn's
1028 --network-interface= switch will now be generated from the
1029 machine name, and thus be stable between multiple invocations
1032 * logind will now automatically remove all IPC objects owned
1033 by a user if she or he fully logs out. This makes sure that
1034 users who are logged out cannot continue to consume IPC
1035 resources. This covers SysV memory, semaphores and message
1036 queues as well as POSIX shared memory and message
1037 queues. Traditionally, SysV and POSIX IPC had no life-cycle
1038 limits. With this functionality, that is corrected. This may
1039 be turned off by using the RemoveIPC= switch of logind.conf.
1041 * The systemd-machine-id-setup and tmpfiles tools gained a
1042 --root= switch to operate on a specific root directory,
1045 * journald can now forward logged messages to the TTYs of all
1046 logged in users ("wall"). This is the default for all
1047 emergency messages now.
1049 * A new tool systemd-journal-remote has been added to stream
1050 journal log messages across the network.
1052 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
1053 controller trees are mounted into it. Note that the
1054 directories mounted beneath it are not read-only. This is a
1055 security measure and is particularly useful because glibc
1056 actually includes a search logic to pick any tmpfs it can
1057 find to implement shm_open() if /dev/shm is not available
1058 (which it might very well be in namespaced setups).
1060 * machinectl gained a new "poweroff" command to cleanly power
1061 down a local OS container.
1063 * The PrivateDevices= unit file setting will now also drop the
1064 CAP_MKNOD capability from the capability bound set, and
1065 imply DevicePolicy=closed.
1067 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
1068 comprehensively on all long-running systemd services where
1069 this is appropriate.
1071 * systemd-udevd will now run in a disassociated mount
1072 namespace. To mount directories from udev rules, make sure to
1073 pull in mount units via SYSTEMD_WANTS properties.
1075 * The kdbus support gained support for uploading policy into
1076 the kernel. sd-bus gained support for creating "monitoring"
1077 connections that can eavesdrop into all bus communication
1078 for debugging purposes.
1080 * Timestamps may now be specified in seconds since the UNIX
1081 epoch Jan 1st, 1970 by specifying "@" followed by the value
1084 * Native tcpwrap support in systemd has been removed. tcpwrap
1085 is old code, not really maintained anymore and has serious
1086 shortcomings, and better options such as firewalls
1087 exist. For setups that require tcpwrap usage, please
1088 consider invoking your socket-activated service via tcpd,
1089 like on traditional inetd.
1091 * A new system.conf configuration option
1092 DefaultTimerAccuracySec= has been added that controls the
1093 default AccuracySec= setting of .timer units.
1095 * Timer units gained a new WakeSystem= switch. If enabled,
1096 timers configured this way will cause the system to resume
1097 from system suspend (if the system supports that, which most
1100 * Timer units gained a new Persistent= switch. If enabled,
1101 timers configured this way will save to disk when they have
1102 been last triggered. This information is then used on next
1103 reboot to possible execute overdue timer events, that
1104 could not take place because the system was powered off.
1105 This enables simple anacron-like behaviour for timer units.
1107 * systemctl's "list-timers" will now also list the time a
1108 timer unit was last triggered in addition to the next time
1109 it will be triggered.
1111 * systemd-networkd will now assign predictable IPv4LL
1112 addresses to its local interfaces.
1114 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
1115 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
1116 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
1117 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
1118 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
1119 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
1120 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
1121 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
1124 -- Berlin, 2014-03-25
1128 * A new unit file setting RestrictAddressFamilies= has been
1129 added to restrict which socket address families unit
1130 processes gain access to. This takes address family names
1131 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1132 attack surface of services via exotic protocol stacks. This
1133 is built on seccomp system call filters.
1135 * Two new unit file settings RuntimeDirectory= and
1136 RuntimeDirectoryMode= have been added that may be used to
1137 manage a per-daemon runtime directories below /run. This is
1138 an alternative for setting up directory permissions with
1139 tmpfiles snippets, and has the advantage that the runtime
1140 directory's lifetime is bound to the daemon runtime and that
1141 the daemon starts up with an empty directory each time. This
1142 is particularly useful when writing services that drop
1143 privileges using the User= or Group= setting.
1145 * The DeviceAllow= unit setting now supports globbing for
1146 matching against device group names.
1148 * The systemd configuration file system.conf gained new
1149 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1150 DefaultMemoryAccounting= to globally turn on/off accounting
1151 for specific resources (cgroups) for all units. These
1152 settings may still be overridden individually in each unit
1155 * systemd-gpt-auto-generator is now able to discover /srv and
1156 root partitions in addition to /home and swap partitions. It
1157 also supports LUKS-encrypted partitions now. With this in
1158 place, automatic discovery of partitions to mount following
1159 the Discoverable Partitions Specification
1160 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1161 is now a lot more complete. This allows booting without
1162 /etc/fstab and without root= on the kernel command line on
1163 systems prepared appropriately.
1165 * systemd-nspawn gained a new --image= switch which allows
1166 booting up disk images and Linux installations on any block
1167 device that follow the Discoverable Partitions Specification
1168 (see above). This means that installations made with
1169 appropriately updated installers may now be started and
1170 deployed using container managers, completely
1171 unmodified. (We hope that libvirt-lxc will add support for
1172 this feature soon, too.)
1174 * systemd-nspawn gained a new --network-macvlan= setting to
1175 set up a private macvlan interface for the
1176 container. Similarly, systemd-networkd gained a new
1177 Kind=macvlan setting in .netdev files.
1179 * systemd-networkd now supports configuring local addresses
1182 * A new tool systemd-network-wait-online has been added to
1183 synchronously wait for network connectivity using
1186 * The sd-bus.h bus API gained a new sd_bus_track object for
1187 tracking the life-cycle of bus peers. Note that sd-bus.h is
1188 still not a public API though (unless you specify
1189 --enable-kdbus on the configure command line, which however
1190 voids your warranty and you get no API stability guarantee).
1192 * The $XDG_RUNTIME_DIR runtime directories for each user are
1193 now individual tmpfs instances, which has the benefit of
1194 introducing separate pools for each user, with individual
1195 size limits, and thus making sure that unprivileged clients
1196 can no longer negatively impact the system or other users by
1197 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
1198 RuntimeDirectorySize= has been introduced that allows
1199 controlling the default size limit for all users. It
1200 defaults to 10% of the available physical memory. This is no
1201 replacement for quotas on tmpfs though (which the kernel
1202 still does not support), as /dev/shm and /tmp are still
1203 shared resources used by both the system and unprivileged
1206 * logind will now automatically turn off automatic suspending
1207 on laptop lid close when more than one display is
1208 connected. This was previously expected to be implemented
1209 individually in desktop environments (such as GNOME),
1210 however has been added to logind now, in order to fix a
1211 boot-time race where a desktop environment might not have
1212 been started yet and thus not been able to take an inhibitor
1213 lock at the time where logind already suspends the system
1214 due to a closed lid.
1216 * logind will now wait at least 30s after each system
1217 suspend/resume cycle, and 3min after system boot before
1218 suspending the system due to a closed laptop lid. This
1219 should give USB docking stations and similar enough time to
1220 be probed and configured after system resume and boot in
1221 order to then act as suspend blocker.
1223 * systemd-run gained a new --property= setting which allows
1224 initialization of resource control properties (and others)
1225 for the created scope or service unit. Example: "systemd-run
1226 --property=BlockIOWeight=10 updatedb" may be used to run
1227 updatedb at a low block IO scheduling weight.
1229 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1230 now also work in --scope mode.
1232 * When systemd is compiled with kdbus support, basic support
1233 for enforced policies is now in place. (Note that enabling
1234 kdbus still voids your warranty and no API compatibility
1237 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1238 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1239 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1240 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1241 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1242 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1243 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1244 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
1245 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
1246 Zbigniew Jędrzejewski-Szmek
1248 -- Berlin, 2014-03-12
1252 * systemd will now relabel /dev after loading the SMACK policy
1253 according to SMACK rules.
1255 * A new unit file option AppArmorProfile= has been added to
1256 set the AppArmor profile for the processes of a unit.
1258 * A new condition check ConditionArchitecture= has been added
1259 to conditionalize units based on the system architecture, as
1260 reported by uname()'s "machine" field.
1262 * systemd-networkd now supports matching on the system
1263 virtualization, architecture, kernel command line, host name
1266 * logind is now a lot more aggressive when suspending the
1267 machine due to a closed laptop lid. Instead of acting only
1268 on the lid close action, it will continuously watch the lid
1269 status and act on it. This is useful for laptops where the
1270 power button is on the outside of the chassis so that it can
1271 be reached without opening the lid (such as the Lenovo
1272 Yoga). On those machines, logind will now immediately
1273 re-suspend the machine if the power button has been
1274 accidentally pressed while the laptop was suspended and in a
1275 backpack or similar.
1277 * logind will now watch SW_DOCK switches and inhibit reaction
1278 to the lid switch if it is pressed. This means that logind
1279 will not suspend the machine anymore if the lid is closed
1280 and the system is docked, if the laptop supports SW_DOCK
1281 notifications via the input layer. Note that ACPI docking
1282 stations do not generate this currently. Also note that this
1283 logic is usually not fully sufficient and Desktop
1284 Environments should take a lid switch inhibitor lock when an
1285 external display is connected, as systemd will not watch
1288 * nspawn will now make use of the devices cgroup controller by
1289 default, and only permit creation of and access to the usual
1290 API device nodes like /dev/null or /dev/random, as well as
1291 access to (but not creation of) the pty devices.
1293 * We will now ship a default .network file for
1294 systemd-networkd that automatically configures DHCP for
1295 network interfaces created by nspawn's --network-veth or
1296 --network-bridge= switches.
1298 * systemd will now understand the usual M, K, G, T suffixes
1299 according to SI conventions (i.e. to the base 1000) when
1300 referring to throughput and hardware metrics. It will stay
1301 with IEC conventions (i.e. to the base 1024) for software
1302 metrics, according to what is customary according to
1303 Wikipedia. We explicitly document which base applies for
1304 each configuration option.
1306 * The DeviceAllow= setting in unit files now supports a syntax
1307 to whitelist an entire group of devices node majors at once,
1308 based on the /proc/devices listing. For example, with the
1309 string "char-pts", it is now possible to whitelist all
1310 current and future pseudo-TTYs at once.
1312 * sd-event learned a new "post" event source. Event sources of
1313 this type are triggered by the dispatching of any event
1314 source of a type that is not "post". This is useful for
1315 implementing clean-up and check event sources that are
1316 triggered by other work being done in the program.
1318 * systemd-networkd is no longer statically enabled, but uses
1319 the usual [Install] sections so that it can be
1320 enabled/disabled using systemctl. It still is enabled by
1323 * When creating a veth interface pair with systemd-nspawn, the
1324 host side will now be prefixed with "vb-" if
1325 --network-bridge= is used, and with "ve-" if --network-veth
1326 is used. This way, it is easy to distinguish these cases on
1327 the host, for example to apply different configuration to
1328 them with systemd-networkd.
1330 * The compatibility libraries for libsystemd-journal.so,
1331 libsystem-id128.so, libsystemd-login.so and
1332 libsystemd-daemon.so do not make use of IFUNC
1333 anymore. Instead, we now build libsystemd.so multiple times
1334 under these alternative names. This means that the footprint
1335 is drastically increased, but given that these are
1336 transitional compatibility libraries, this should not matter
1337 much. This change has been made necessary to support the ARM
1338 platform for these compatibility libraries, as the ARM
1339 toolchain is not really at the same level as the toolchain
1340 for other architectures like x86 and does not support
1341 IFUNC. Please make sure to use --enable-compat-libs only
1342 during a transitional period!
1344 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
1345 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1346 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
1347 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
1348 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
1349 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
1350 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
1351 Zbigniew Jędrzejewski-Szmek
1353 -- Berlin, 2014-02-24
1357 * A new component "systemd-networkd" has been added that can
1358 be used to configure local network interfaces statically or
1359 via DHCP. It is capable of bringing up bridges, VLANs, and
1360 bonding. Currently, no hook-ups for interactive network
1361 configuration are provided. Use this for your initrd,
1362 container, embedded, or server setup if you need a simple,
1363 yet powerful, network configuration solution. This
1364 configuration subsystem is quite nifty, as it allows wildcard
1365 hotplug matching in interfaces. For example, with a single
1366 configuration snippet, you can configure that all Ethernet
1367 interfaces showing up are automatically added to a bridge,
1368 or similar. It supports link-sensing and more.
1370 * A new tool "systemd-socket-proxyd" has been added which can
1371 act as a bidirectional proxy for TCP sockets. This is
1372 useful for adding socket activation support to services that
1373 do not actually support socket activation, including virtual
1374 machines and the like.
1376 * Add a new tool to save/restore rfkill state on
1379 * Save/restore state of keyboard backlights in addition to
1380 display backlights on shutdown/boot.
1382 * udev learned a new SECLABEL{} construct to label device
1383 nodes with a specific security label when they appear. For
1384 now, only SECLABEL{selinux} is supported, but the syntax is
1385 prepared for additional security frameworks.
1387 * udev gained a new scheme to configure link-level attributes
1388 from files in /etc/systemd/network/*.link. These files can
1389 match against MAC address, device path, driver name and type,
1390 and will apply attributes like the naming policy, link speed,
1391 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
1392 address assignment policy (randomized, ...).
1394 * The configuration of network interface naming rules for
1395 "permanent interface names" has changed: a new NamePolicy=
1396 setting in the [Link] section of .link files determines the
1397 priority of possible naming schemes (onboard, slot, mac,
1398 path). The default value of this setting is determined by
1399 /usr/lib/net/links/99-default.link. Old
1400 80-net-name-slot.rules udev configuration file has been
1401 removed, so local configuration overriding this file should
1402 be adapated to override 99-default.link instead.
1404 * When the User= switch is used in a unit file, also
1405 initialize $SHELL= based on the user database entry.
1407 * systemd no longer depends on libdbus. All communication is
1408 now done with sd-bus, systemd's low-level bus library
1411 * kdbus support has been added to PID 1 itself. When kdbus is
1412 enabled, this causes PID 1 to set up the system bus and
1413 enable support for a new ".busname" unit type that
1414 encapsulates bus name activation on kdbus. It works a little
1415 bit like ".socket" units, except for bus names. A new
1416 generator has been added that converts classic dbus1 service
1417 activation files automatically into native systemd .busname
1420 * sd-bus: add a light-weight vtable implementation that allows
1421 defining objects on the bus with a simple static const
1422 vtable array of its methods, signals and properties.
1424 * systemd will not generate or install static dbus
1425 introspection data anymore to /usr/share/dbus-1/interfaces,
1426 as the precise format of these files is unclear, and
1427 nothing makes use of it.
1429 * A proxy daemon is now provided to proxy clients connecting
1430 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1431 compatibility with classic D-Bus.
1433 * A bus driver implementation has been added that supports the
1434 classic D-Bus bus driver calls on kdbus, also for
1435 compatibility purposes.
1437 * A new API "sd-event.h" has been added that implements a
1438 minimal event loop API built around epoll. It provides a
1439 couple of features that direct epoll usage is lacking:
1440 prioritization of events, scales to large numbers of timer
1441 events, per-event timer slack (accuracy), system-wide
1442 coalescing of timer events, exit handlers, watchdog
1443 supervision support using systemd's sd_notify() API, child
1446 * A new API "sd-rntl.h" has been added that provides an API
1447 around the route netlink interface of the kernel, similar in
1448 style to "sd-bus.h".
1450 * A new API "sd-dhcp-client.h" has been added that provides a
1451 small DHCPv4 client-side implementation. This is used by
1454 * There is a new kernel command line option
1455 "systemd.restore_state=0|1". When set to "0", none of the
1456 systemd tools will restore saved runtime state to hardware
1457 devices. More specifically, the rfkill and backlight states
1460 * The FsckPassNo= compatibility option in mount/service units
1461 has been removed. The fstab generator will now add the
1462 necessary dependencies automatically, and does not require
1463 PID1's support for that anymore.
1465 * journalctl gained a new switch, --list-boots, that lists
1466 recent boots with their times and boot IDs.
1468 * The various tools like systemctl, loginctl, timedatectl,
1469 busctl, systemd-run, ... have gained a new switch "-M" to
1470 connect to a specific, local OS container (as direct
1471 connection, without requiring SSH). This works on any
1472 container that is registered with machined, such as those
1473 created by libvirt-lxc or nspawn.
1475 * systemd-run and systemd-analyze also gained support for "-H"
1476 to connect to remote hosts via SSH. This is particularly
1477 useful for systemd-run because it enables queuing of jobs
1478 onto remote systems.
1480 * machinectl gained a new command "login" to open a getty
1481 login in any local container. This works with any container
1482 that is registered with machined (such as those created by
1483 libvirt-lxc or nspawn), and which runs systemd inside.
1485 * machinectl gained a new "reboot" command that may be used to
1486 trigger a reboot on a specific container that is registered
1487 with machined. This works on any container that runs an init
1488 system of some kind.
1490 * systemctl gained a new "list-timers" command to print a nice
1491 listing of installed timer units with the times they elapse
1494 * Alternative reboot() parameters may now be specified on the
1495 "systemctl reboot" command line and are passed to the
1496 reboot() system call.
1498 * systemctl gained a new --job-mode= switch to configure the
1499 mode to queue a job with. This is a more generic version of
1500 --fail, --irreversible, and --ignore-dependencies, which are
1501 still available but not advertised anymore.
1503 * /etc/systemd/system.conf gained new settings to configure
1504 various default timeouts of units, as well as the default
1505 start limit interval and burst. These may still be overridden
1508 * PID1 will now export on the bus profile data of the security
1509 policy upload process (such as the SELinux policy upload to
1512 * journald: when forwarding logs to the console, include
1513 timestamps (following the setting in
1514 /sys/module/printk/parameters/time).
1516 * OnCalendar= in timer units now understands the special
1517 strings "yearly" and "annually". (Both are equivalent)
1519 * The accuracy of timer units is now configurable with the new
1520 AccuracySec= setting. It defaults to 1min.
1522 * A new dependency type JoinsNamespaceOf= has been added that
1523 allows running two services within the same /tmp and network
1524 namespace, if PrivateNetwork= or PrivateTmp= are used.
1526 * A new command "cat" has been added to systemctl. It outputs
1527 the original unit file of a unit, and concatenates the
1528 contents of additional "drop-in" unit file snippets, so that
1529 the full configuration is shown.
1531 * systemctl now supports globbing on the various "list-xyz"
1532 commands, like "list-units" or "list-sockets", as well as on
1533 those commands which take multiple unit names.
1535 * journalctl's --unit= switch gained support for globbing.
1537 * All systemd daemons now make use of the watchdog logic so
1538 that systemd automatically notices when they hang.
1540 * If the $container_ttys environment variable is set,
1541 getty-generator will automatically spawn a getty for each
1542 listed tty. This is useful for container managers to request
1543 login gettys to be spawned on as many ttys as needed.
1545 * %h, %s, %U specifier support is not available anymore when
1546 used in unit files for PID 1. This is because NSS calls are
1547 not safe from PID 1. They stay available for --user
1548 instances of systemd, and as special case for the root user.
1550 * loginctl gained a new "--no-legend" switch to turn off output
1553 * The "sd-login.h" API gained three new calls:
1554 sd_session_is_remote(), sd_session_get_remote_user(),
1555 sd_session_get_remote_host() to query information about
1558 * The udev hardware database now also carries vendor/product
1559 information of SDIO devices.
1561 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
1562 determine whether watchdog notifications are requested by
1565 * Socket-activated per-connection services now include a
1566 short description of the connection parameters in the
1569 * tmpfiles gained a new "--boot" option. When this is not used,
1570 only lines where the command character is not suffixed with
1571 "!" are executed. When this option is specified, those
1572 options are executed too. This partitions tmpfiles
1573 directives into those that can be safely executed at any
1574 time, and those which should be run only at boot (for
1575 example, a line that creates /run/nologin).
1577 * A new API "sd-resolve.h" has been added which provides a simple
1578 asynchronous wrapper around glibc NSS host name resolution
1579 calls, such as getaddrinfo(). In contrast to glibc's
1580 getaddrinfo_a(), it does not use signals. In contrast to most
1581 other asynchronous name resolution libraries, this one does
1582 not reimplement DNS, but reuses NSS, so that alternate
1583 host name resolution systems continue to work, such as mDNS,
1584 LDAP, etc. This API is based on libasyncns, but it has been
1585 cleaned up for inclusion in systemd.
1587 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
1588 "sd-daemon.h" are no longer found in individual libraries
1589 libsystemd-journal.so, libsystemd-login.so,
1590 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
1591 merged them into a single library, libsystemd.so, which
1592 provides all symbols. The reason for this is cyclic
1593 dependencies, as these libraries tend to use each other's
1594 symbols. So far, we have managed to workaround that by linking
1595 a copy of a good part of our code into each of these
1596 libraries again and again, which, however, makes certain
1597 things hard to do, like sharing static variables. Also, it
1598 substantially increases footprint. With this change, there
1599 is only one library for the basic APIs systemd
1600 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
1601 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
1602 library as well, however are subject to the --enable-kdbus
1603 switch (see below). Note that "sd-dhcp-client.h" is not part
1604 of this library (this is because it only consumes, never
1605 provides, services of/to other APIs). To make the transition
1606 easy from the separate libraries to the unified one, we
1607 provide the --enable-compat-libs compile-time switch which
1608 will generate stub libraries that are compatible with the
1609 old ones but redirect all calls to the new one.
1611 * All of the kdbus logic and the new APIs "sd-bus.h",
1612 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
1613 and "sd-utf8.h" are compile-time optional via the
1614 "--enable-kdbus" switch, and they are not compiled in by
1615 default. To make use of kdbus, you have to explicitly enable
1616 the switch. Note however, that neither the kernel nor the
1617 userspace API for all of this is considered stable yet. We
1618 want to maintain the freedom to still change the APIs for
1619 now. By specifying this build-time switch, you acknowledge
1620 that you are aware of the instability of the current
1623 * Also, note that while kdbus is pretty much complete,
1624 it lacks one thing: proper policy support. This means you
1625 can build a fully working system with all features; however,
1626 it will be highly insecure. Policy support will be added in
1627 one of the next releases, at the same time that we will
1628 declare the APIs stable.
1630 * When the kernel command-line argument "kdbus" is specified,
1631 systemd will automatically load the kdbus.ko kernel module. At
1632 this stage of development, it is only useful for testing kdbus
1633 and should not be used in production. Note: if "--enable-kdbus"
1634 is specified, and the kdbus.ko kernel module is available, and
1635 "kdbus" is added to the kernel command line, the entire system
1636 runs with kdbus instead of dbus-daemon, with the above mentioned
1637 problem of missing the system policy enforcement. Also a future
1638 version of kdbus.ko or a newer systemd will not be compatible with
1639 each other, and will unlikely be able to boot the machine if only
1640 one of them is updated.
1642 * systemctl gained a new "import-environment" command which
1643 uploads the caller's environment (or parts thereof) into the
1644 service manager so that it is inherited by services started
1645 by the manager. This is useful to upload variables like
1646 $DISPLAY into the user service manager.
1648 * A new PrivateDevices= switch has been added to service units
1649 which allows running a service with a namespaced /dev
1650 directory that does not contain any device nodes for
1651 physical devices. More specifically, it only includes devices
1652 such as /dev/null, /dev/urandom, and /dev/zero which are API
1655 * logind has been extended to support behaviour like VT
1656 switching on seats that do not support a VT. This makes
1657 multi-session available on seats that are not the first seat
1658 (seat0), and on systems where kernel support for VTs has
1659 been disabled at compile-time.
1661 * If a process holds a delay lock for system sleep or shutdown
1662 and fails to release it in time, we will now log its
1663 identity. This makes it easier to identify processes that
1664 cause slow suspends or power-offs.
1666 * When parsing /etc/crypttab, support for a new key-slot=
1667 option as supported by Debian is added. It allows indicating
1668 which LUKS slot to use on disk, speeding up key loading.
1670 * The sd_journald_sendv() API call has been checked and
1671 officially declared to be async-signal-safe so that it may
1672 be invoked from signal handlers for logging purposes.
1674 * Boot-time status output is now enabled automatically after a
1675 short timeout if boot does not progress, in order to give
1676 the user an indication what she or he is waiting for.
1678 * The boot-time output has been improved to show how much time
1679 remains until jobs expire.
1681 * The KillMode= switch in service units gained a new possible
1682 value "mixed". If set, and the unit is shut down, then the
1683 initial SIGTERM signal is sent only to the main daemon
1684 process, while the following SIGKILL signal is sent to
1685 all remaining processes of the service.
1687 * When a scope unit is registered, a new property "Controller"
1688 may be set. If set to a valid bus name, systemd will send a
1689 RequestStop() signal to this name when it would like to shut
1690 down the scope. This may be used to hook manager logic into
1691 the shutdown logic of scope units. Also, scope units may now
1692 be put in a special "abandoned" state, in which case the
1693 manager process which created them takes no further
1694 responsibilities for it.
1696 * When reading unit files, systemd will now verify
1697 the access mode of these files, and warn about certain
1698 suspicious combinations. This has been added to make it
1699 easier to track down packaging bugs where unit files are
1700 marked executable or world-writable.
1702 * systemd-nspawn gained a new "--setenv=" switch to set
1703 container-wide environment variables. The similar option in
1704 systemd-activate was renamed from "--environment=" to
1705 "--setenv=" for consistency.
1707 * systemd-nspawn has been updated to create a new kdbus domain
1708 for each container that is invoked, thus allowing each
1709 container to have its own set of system and user buses,
1710 independent of the host.
1712 * systemd-nspawn gained a new --drop-capability= switch to run
1713 the container with less capabilities than the default. Both
1714 --drop-capability= and --capability= now take the special
1715 string "all" for dropping or keeping all capabilities.
1717 * systemd-nspawn gained new switches for executing containers
1718 with specific SELinux labels set.
1720 * systemd-nspawn gained a new --quiet switch to not generate
1721 any additional output but the container's own console
1724 * systemd-nspawn gained a new --share-system switch to run a
1725 container without PID namespacing enabled.
1727 * systemd-nspawn gained a new --register= switch to control
1728 whether the container is registered with systemd-machined or
1729 not. This is useful for containers that do not run full
1730 OS images, but only specific apps.
1732 * systemd-nspawn gained a new --keep-unit which may be used
1733 when invoked as the only program from a service unit, and
1734 results in registration of the unit service itself in
1735 systemd-machined, instead of a newly opened scope unit.
1737 * systemd-nspawn gained a new --network-interface= switch for
1738 moving arbitrary interfaces to the container. The new
1739 --network-veth switch creates a virtual Ethernet connection
1740 between host and container. The new --network-bridge=
1741 switch then allows assigning the host side of this virtual
1742 Ethernet connection to a bridge device.
1744 * systemd-nspawn gained a new --personality= switch for
1745 setting the kernel personality for the container. This is
1746 useful when running a 32-bit container on a 64-bit host. A
1747 similar option Personality= is now also available for service
1750 * logind will now also track a "Desktop" identifier for each
1751 session which encodes the desktop environment of it. This is
1752 useful for desktop environments that want to identify
1753 multiple running sessions of itself easily.
1755 * A new SELinuxContext= setting for service units has been
1756 added that allows setting a specific SELinux execution
1757 context for a service.
1759 * Most systemd client tools will now honour $SYSTEMD_LESS for
1760 settings of the "less" pager. By default, these tools will
1761 override $LESS to allow certain operations to work, such as
1762 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1763 influence this logic.
1765 * systemd's "seccomp" hook-up has been changed to make use of
1766 the libseccomp library instead of using its own
1767 implementation. This has benefits for portability among
1770 * For usage together with SystemCallFilter=, a new
1771 SystemCallErrorNumber= setting has been introduced that
1772 allows configuration of a system error number to be returned
1773 on filtered system calls, instead of immediately killing the
1774 process. Also, SystemCallArchitectures= has been added to
1775 limit access to system calls of a particular architecture
1776 (in order to turn off support for unused secondary
1777 architectures). There is also a global
1778 SystemCallArchitectures= setting in system.conf now to turn
1779 off support for non-native system calls system-wide.
1781 * systemd requires a kernel with a working name_to_handle_at(),
1782 please see the kernel config requirements in the README file.
1784 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1785 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1786 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1787 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1788 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1789 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1790 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1791 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1792 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1793 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1794 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1795 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1796 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1797 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1798 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1799 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1800 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1801 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1802 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1803 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1804 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1805 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1806 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1807 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1809 -- Berlin, 2014-02-20
1813 * logind has gained support for facilitating privileged input
1814 and drm device access for unprivileged clients. This work is
1815 useful to allow Wayland display servers (and similar
1816 programs, such as kmscon) to run under the user's ID and
1817 access input and drm devices which are normally
1818 protected. When this is used (and the kernel is new enough)
1819 logind will "mute" IO on the file descriptors passed to
1820 Wayland as long as it is in the background and "unmute" it
1821 if it returns into the foreground. This allows secure
1822 session switching without allowing background sessions to
1823 eavesdrop on input and display data. This also introduces
1824 session switching support if VT support is turned off in the
1825 kernel, and on seats that are not seat0.
1827 * A new kernel command line option luks.options= is understood
1828 now which allows specifying LUKS options for usage for LUKS
1829 encrypted partitions specified with luks.uuid=.
1831 * tmpfiles.d(5) snippets may now use specifier expansion in
1832 path names. More specifically %m, %b, %H, %v, are now
1833 replaced by the local machine id, boot id, hostname, and
1834 kernel version number.
1836 * A new tmpfiles.d(5) command "m" has been introduced which
1837 may be used to change the owner/group/access mode of a file
1838 or directory if it exists, but do nothing if it does not.
1840 * This release removes high-level support for the
1841 MemorySoftLimit= cgroup setting. The underlying kernel
1842 cgroup attribute memory.soft_limit= is currently badly
1843 designed and likely to be removed from the kernel API in its
1844 current form, hence we should not expose it for now.
1846 * The memory.use_hierarchy cgroup attribute is now enabled for
1847 all cgroups systemd creates in the memory cgroup
1848 hierarchy. This option is likely to be come the built-in
1849 default in the kernel anyway, and the non-hierarchial mode
1850 never made much sense in the intrinsically hierarchial
1853 * A new field _SYSTEMD_SLICE= is logged along with all journal
1854 messages containing the slice a message was generated
1855 from. This is useful to allow easy per-customer filtering of
1856 logs among other things.
1858 * systemd-journald will no longer adjust the group of journal
1859 files it creates to the "systemd-journal" group. Instead we
1860 rely on the journal directory to be owned by the
1861 "systemd-journal" group, and its setgid bit set, so that the
1862 kernel file system layer will automatically enforce that
1863 journal files inherit this group assignment. The reason for
1864 this change is that we cannot allow NSS look-ups from
1865 journald which would be necessary to resolve
1866 "systemd-journal" to a numeric GID, because this might
1867 create deadlocks if NSS involves synchronous queries to
1868 other daemons (such as nscd, or sssd) which in turn are
1869 logging clients of journald and might block on it, which
1870 would then dead lock. A tmpfiles.d(5) snippet included in
1871 systemd will make sure the setgid bit and group are
1872 properly set on the journal directory if it exists on every
1873 boot. However, we recommend adjusting it manually after
1874 upgrades too (or from RPM scriptlets), so that the change is
1875 not delayed until next reboot.
1877 * Backlight and random seed files in /var/lib/ have moved into
1878 the /var/lib/systemd/ directory, in order to centralize all
1879 systemd generated files in one directory.
1881 * Boot time performance measurements (as displayed by
1882 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1883 performance information if that's available to determine how
1884 much time BIOS and boot loader initialization required. With
1885 a sufficiently new BIOS you hence no longer need to boot
1886 with Gummiboot to get access to such information.
1888 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1889 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1890 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1891 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1892 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1893 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1894 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1896 -- Berlin, 2013-10-02
1900 * The Restart= option for services now understands a new
1901 on-watchdog setting, which will restart the service
1902 automatically if the service stops sending out watchdog keep
1903 alive messages (as configured with WatchdogSec=).
1905 * The getty generator (which is responsible for bringing up a
1906 getty on configured serial consoles) will no longer only
1907 start a getty on the primary kernel console but on all
1908 others, too. This makes the order in which console= is
1909 specified on the kernel command line less important.
1911 * libsystemd-logind gained a new sd_session_get_vt() call to
1912 retrieve the VT number of a session.
1914 * If the option "tries=0" is set for an entry of /etc/crypttab
1915 its passphrase is queried indefinitely instead of any
1916 maximum number of tries.
1918 * If a service with a configure PID file terminates its PID
1919 file will now be removed automatically if it still exists
1920 afterwards. This should put an end to stale PID files.
1922 * systemd-run will now also take relative binary path names
1923 for execution and no longer insists on absolute paths.
1925 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1926 paths that are optionally prefixed with "-" to indicate that
1927 it should not be considered a failure if they do not exist.
1929 * journalctl -o (and similar commands) now understands a new
1930 output mode "short-precise", it is similar to "short" but
1931 shows timestamps with usec accuracy.
1933 * The option "discard" (as known from Debian) is now
1934 synonymous to "allow-discards" in /etc/crypttab. In fact,
1935 "discard" is preferred now (since it is easier to remember
1938 * Some licensing clean-ups were made, so that more code is now
1939 LGPL-2.1 licensed than before.
1941 * A minimal tool to save/restore the display backlight
1942 brightness across reboots has been added. It will store the
1943 backlight setting as late as possible at shutdown, and
1944 restore it as early as possible during reboot.
1946 * A logic to automatically discover and enable home and swap
1947 partitions on GPT disks has been added. With this in place
1948 /etc/fstab becomes optional for many setups as systemd can
1949 discover certain partitions located on the root disk
1950 automatically. Home partitions are recognized under their
1951 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1952 partitions are recognized under their GPT type ID
1953 0657fd6da4ab43c484e50933c84b4f4f.
1955 * systemd will no longer pass any environment from the kernel
1956 or initrd to system services. If you want to set an
1957 environment for all services, do so via the kernel command
1958 line systemd.setenv= assignment.
1960 * The systemd-sysctl tool no longer natively reads the file
1961 /etc/sysctl.conf. If desired, the file should be symlinked
1962 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1963 legacy support by a symlink rather than built-in code, it
1964 also makes the otherwise hidden order of application of the
1965 different files visible. (Note that this partly reverts to a
1966 pre-198 application order of sysctl knobs!)
1968 * The "systemctl set-log-level" and "systemctl dump" commands
1969 have been moved to systemd-analyze.
1971 * systemd-run learned the new --remain-after-exit switch,
1972 which causes the scope unit not to be cleaned up
1973 automatically after the process terminated.
1975 * tmpfiles learned a new --exclude-prefix= switch to exclude
1976 certain paths from operation.
1978 * journald will now automatically flush all messages to disk
1979 as soon as a message at the log level CRIT, ALERT or EMERG
1982 Contributions from: Andrew Cook, Brandon Philips, Christian
1983 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1984 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1985 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1986 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1987 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1988 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1989 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1990 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1991 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1992 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1993 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1994 William Giokas, Zbigniew Jędrzejewski-Szmek
1996 -- Berlin, 2013-09-13
2000 * The documentation has been updated to cover the various new
2001 concepts introduced with 205.
2003 * Unit files now understand the new %v specifier which
2004 resolves to the kernel version string as returned by "uname
2007 * systemctl now supports filtering the unit list output by
2008 load state, active state and sub state, using the new
2011 * "systemctl status" will now show the results of the
2012 condition checks (like ConditionPathExists= and similar) of
2013 the last start attempts of the unit. They are also logged to
2016 * "journalctl -b" may now be used to look for boot output of a
2017 specific boot. Try "journalctl -b -1" for the previous boot,
2018 but the syntax is substantially more powerful.
2020 * "journalctl --show-cursor" has been added which prints the
2021 cursor string the last shown log line. This may then be used
2022 with the new "journalctl --after-cursor=" switch to continue
2023 browsing logs from that point on.
2025 * "journalctl --force" may now be used to force regeneration
2028 * Creation of "dead" device nodes has been moved from udev
2029 into kmod and tmpfiles. Previously, udev would read the kmod
2030 databases to pre-generate dead device nodes based on meta
2031 information contained in kernel modules, so that these would
2032 be auto-loaded on access rather then at boot. As this
2033 does not really have much to do with the exposing actual
2034 kernel devices to userspace this has always been slightly
2035 alien in the udev codebase. Following the new scheme kmod
2036 will now generate a runtime snippet for tmpfiles from the
2037 module meta information and it now is tmpfiles' job to the
2038 create the nodes. This also allows overriding access and
2039 other parameters for the nodes using the usual tmpfiles
2040 facilities. As side effect this allows us to remove the
2041 CAP_SYS_MKNOD capability bit from udevd entirely.
2043 * logind's device ACLs may now be applied to these "dead"
2044 devices nodes too, thus finally allowing managed access to
2045 devices such as /dev/snd/sequencer whithout loading the
2046 backing module right-away.
2048 * A new RPM macro has been added that may be used to apply
2049 tmpfiles configuration during package installation.
2051 * systemd-detect-virt and ConditionVirtualization= now can
2052 detect User-Mode-Linux machines (UML).
2054 * journald will now implicitly log the effective capabilities
2055 set of processes in the message metadata.
2057 * systemd-cryptsetup has gained support for TrueCrypt volumes.
2059 * The initrd interface has been simplified (more specifically,
2060 support for passing performance data via environment
2061 variables and fsck results via files in /run has been
2062 removed). These features were non-essential, and are
2063 nowadays available in a much nicer way by having systemd in
2064 the initrd serialize its state and have the hosts systemd
2065 deserialize it again.
2067 * The udev "keymap" data files and tools to apply keyboard
2068 specific mappings of scan to key codes, and force-release
2069 scan code lists have been entirely replaced by a udev
2070 "keyboard" builtin and a hwdb data file.
2072 * systemd will now honour the kernel's "quiet" command line
2073 argument also during late shutdown, resulting in a
2074 completely silent shutdown when used.
2076 * There's now an option to control the SO_REUSEPORT socket
2077 option in .socket units.
2079 * Instance units will now automatically get a per-template
2080 subslice of system.slice unless something else is explicitly
2081 configured. For example, instances of sshd@.service will now
2082 implicitly be placed in system-sshd.slice rather than
2083 system.slice as before.
2085 * Test coverage support may now be enabled at build time.
2087 Contributions from: Dave Reisner, Frederic Crozat, Harald
2088 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
2089 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
2090 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
2091 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
2092 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
2093 Giokas, Zbigniew Jędrzejewski-Szmek
2095 -- Berlin, 2013-07-23
2099 * Two new unit types have been introduced:
2101 Scope units are very similar to service units, however, are
2102 created out of pre-existing processes -- instead of PID 1
2103 forking off the processes. By using scope units it is
2104 possible for system services and applications to group their
2105 own child processes (worker processes) in a powerful way
2106 which then maybe used to organize them, or kill them
2107 together, or apply resource limits on them.
2109 Slice units may be used to partition system resources in an
2110 hierarchial fashion and then assign other units to them. By
2111 default there are now three slices: system.slice (for all
2112 system services), user.slice (for all user sessions),
2113 machine.slice (for VMs and containers).
2115 Slices and scopes have been introduced primarily in
2116 context of the work to move cgroup handling to a
2117 single-writer scheme, where only PID 1
2118 creates/removes/manages cgroups.
2120 * There's a new concept of "transient" units. In contrast to
2121 normal units these units are created via an API at runtime,
2122 not from configuration from disk. More specifically this
2123 means it is now possible to run arbitrary programs as
2124 independent services, with all execution parameters passed
2125 in via bus APIs rather than read from disk. Transient units
2126 make systemd substantially more dynamic then it ever was,
2127 and useful as a general batch manager.
2129 * logind has been updated to make use of scope and slice units
2130 for managing user sessions. As a user logs in he will get
2131 his own private slice unit, to which all sessions are added
2132 as scope units. We also added support for automatically
2133 adding an instance of user@.service for the user into the
2134 slice. Effectively logind will no longer create cgroup
2135 hierarchies on its own now, it will defer entirely to PID 1
2136 for this by means of scope, service and slice units. Since
2137 user sessions this way become entities managed by PID 1
2138 the output of "systemctl" is now a lot more comprehensive.
2140 * A new mini-daemon "systemd-machined" has been added which
2141 may be used by virtualization managers to register local
2142 VMs/containers. nspawn has been updated accordingly, and
2143 libvirt will be updated shortly. machined will collect a bit
2144 of meta information about the VMs/containers, and assign
2145 them their own scope unit (see above). The collected
2146 meta-data is then made available via the "machinectl" tool,
2147 and exposed in "ps" and similar tools. machined/machinectl
2148 is compile-time optional.
2150 * As discussed earlier, the low-level cgroup configuration
2151 options ControlGroup=, ControlGroupModify=,
2152 ControlGroupPersistent=, ControlGroupAttribute= have been
2153 removed. Please use high-level attribute settings instead as
2154 well as slice units.
2156 * A new bus call SetUnitProperties() has been added to alter
2157 various runtime parameters of a unit. This is primarily
2158 useful to alter cgroup parameters dynamically in a nice way,
2159 but will be extended later on to make more properties
2160 modifiable at runtime. systemctl gained a new set-properties
2161 command that wraps this call.
2163 * A new tool "systemd-run" has been added which can be used to
2164 run arbitrary command lines as transient services or scopes,
2165 while configuring a number of settings via the command
2166 line. This tool is currently very basic, however already
2167 very useful. We plan to extend this tool to even allow
2168 queuing of execution jobs with time triggers from the
2169 command line, similar in fashion to "at".
2171 * nspawn will now inform the user explicitly that kernels with
2172 audit enabled break containers, and suggest the user to turn
2175 * Support for detecting the IMA and AppArmor security
2176 frameworks with ConditionSecurity= has been added.
2178 * journalctl gained a new "-k" switch for showing only kernel
2179 messages, mimicking dmesg output; in addition to "--user"
2180 and "--system" switches for showing only user's own logs
2183 * systemd-delta can now show information about drop-in
2184 snippets extending unit files.
2186 * libsystemd-bus has been substantially updated but is still
2187 not available as public API.
2189 * systemd will now look for the "debug" argument on the kernel
2190 command line and enable debug logging, similar to what
2191 "systemd.log_level=debug" already did before.
2193 * "systemctl set-default", "systemctl get-default" has been
2194 added to configure the default.target symlink, which
2195 controls what to boot into by default.
2197 * "systemctl set-log-level" has been added as a convenient
2198 way to raise and lower systemd logging threshold.
2200 * "systemd-analyze plot" will now show the time the various
2201 generators needed for execution, as well as information
2202 about the unit file loading.
2204 * libsystemd-journal gained a new sd_journal_open_files() call
2205 for opening specific journal files. journactl also gained a
2206 new switch to expose this new functionality. Previously we
2207 only supported opening all files from a directory, or all
2208 files from the system, as opening individual files only is
2209 racy due to journal file rotation.
2211 * systemd gained the new DefaultEnvironment= setting in
2212 /etc/systemd/system.conf to set environment variables for
2215 * If a privileged process logs a journal message with the
2216 OBJECT_PID= field set, then journald will automatically
2217 augment this with additional OBJECT_UID=, OBJECT_GID=,
2218 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2219 system services want to log events about specific client
2220 processes. journactl/systemctl has been updated to make use
2221 of this information if all log messages regarding a specific
2224 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2225 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2226 Reisner, David Coppa, David King, David Strauss, Eelco
2227 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2228 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2229 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2230 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2231 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2232 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2233 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2234 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2235 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2236 Łukasz Stelmach, 장동준
2240 * The Python bindings gained some minimal support for the APIs
2241 exposed by libsystemd-logind.
2243 * ConditionSecurity= gained support for detecting SMACK. Since
2244 this condition already supports SELinux and AppArmor we only
2245 miss IMA for this. Patches welcome!
2247 Contributions from: Karol Lewandowski, Lennart Poettering,
2248 Zbigniew Jędrzejewski-Szmek
2252 * systemd-nspawn will now create /etc/resolv.conf if
2253 necessary, before bind-mounting the host's file onto it.
2255 * systemd-nspawn will now store meta information about a
2256 container on the container's cgroup as extended attribute
2257 fields, including the root directory.
2259 * The cgroup hierarchy has been reworked in many ways. All
2260 objects any of the components systemd creates in the cgroup
2261 tree are now suffixed. More specifically, user sessions are
2262 now placed in cgroups suffixed with ".session", users in
2263 cgroups suffixed with ".user", and nspawn containers in
2264 cgroups suffixed with ".nspawn". Furthermore, all cgroup
2265 names are now escaped in a simple scheme to avoid collision
2266 of userspace object names with kernel filenames. This work
2267 is preparation for making these objects relocatable in the
2268 cgroup tree, in order to allow easy resource partitioning of
2269 these objects without causing naming conflicts.
2271 * systemctl list-dependencies gained the new switches
2272 --plain, --reverse, --after and --before.
2274 * systemd-inhibit now shows the process name of processes that
2275 have taken an inhibitor lock.
2277 * nss-myhostname will now also resolve "localhost"
2278 implicitly. This makes /etc/hosts an optional file and
2279 nicely handles that on IPv6 ::1 maps to both "localhost" and
2282 * libsystemd-logind.so gained a new call
2283 sd_get_machine_names() to enumerate running containers and
2284 VMs (currently only supported by very new libvirt and
2285 nspawn). sd_login_monitor can now be used to watch
2286 VMs/containers coming and going.
2288 * .include is not allowed recursively anymore, and only in
2289 unit files. Usually it is better to use drop-in snippets in
2290 .d/*.conf anyway, as introduced with systemd 198.
2292 * systemd-analyze gained a new "critical-chain" command that
2293 determines the slowest chain of units run during system
2294 boot-up. It is very useful for tracking down where
2295 optimizing boot time is the most beneficial.
2297 * systemd will no longer allow manipulating service paths in
2298 the name=systemd:/system cgroup tree using ControlGroup= in
2299 units. (But is still fine with it in all other dirs.)
2301 * There's a new systemd-nspawn@.service service file that may
2302 be used to easily run nspawn containers as system
2303 services. With the container's root directory in
2304 /var/lib/container/foobar it is now sufficient to run
2305 "systemctl start systemd-nspawn@foobar.service" to boot it.
2307 * systemd-cgls gained a new parameter "--machine" to list only
2308 the processes within a certain container.
2310 * ConditionSecurity= now can check for "apparmor". We still
2311 are lacking checks for SMACK and IMA for this condition
2312 check though. Patches welcome!
2314 * A new configuration file /etc/systemd/sleep.conf has been
2315 added that may be used to configure which kernel operation
2316 systemd is supposed to execute when "suspend", "hibernate"
2317 or "hybrid-sleep" is requested. This makes the new kernel
2318 "freeze" state accessible to the user.
2320 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
2321 the passed argument if applicable.
2323 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
2324 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
2325 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
2326 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
2327 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
2328 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
2329 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
2334 * The output of 'systemctl list-jobs' got some polishing. The
2335 '--type=' argument may now be passed more than once. A new
2336 command 'systemctl list-sockets' has been added which shows
2337 a list of kernel sockets systemd is listening on with the
2338 socket units they belong to, plus the units these socket
2341 * The experimental libsystemd-bus library got substantial
2342 updates to work in conjunction with the (also experimental)
2343 kdbus kernel project. It works well enough to exchange
2344 messages with some sophistication. Note that kdbus is not
2345 ready yet, and the library is mostly an elaborate test case
2346 for now, and not installable.
2348 * systemd gained a new unit 'systemd-static-nodes.service'
2349 that generates static device nodes earlier during boot, and
2350 can run in conjunction with udev.
2352 * libsystemd-login gained a new call sd_pid_get_user_unit()
2353 to retrieve the user systemd unit a process is running
2354 in. This is useful for systems where systemd is used as
2357 * systemd-nspawn now places all containers in the new /machine
2358 top-level cgroup directory in the name=systemd
2359 hierarchy. libvirt will soon do the same, so that we get a
2360 uniform separation of /system, /user and /machine for system
2361 services, user processes and containers/virtual
2362 machines. This new cgroup hierarchy is also useful to stick
2363 stable names to specific container instances, which can be
2364 recognized later this way (this name may be controlled
2365 via systemd-nspawn's new -M switch). libsystemd-login also
2366 gained a new call sd_pid_get_machine_name() to retrieve the
2367 name of the container/VM a specific process belongs to.
2369 * bootchart can now store its data in the journal.
2371 * libsystemd-journal gained a new call
2372 sd_journal_add_conjunction() for AND expressions to the
2373 matching logic. This can be used to express more complex
2374 logical expressions.
2376 * journactl can now take multiple --unit= and --user-unit=
2379 * The cryptsetup logic now understands the "luks.key=" kernel
2380 command line switch for specifying a file to read the
2381 decryption key from. Also, if a configured key file is not
2382 found the tool will now automatically fall back to prompting
2385 * Python systemd.journal module was updated to wrap recently
2386 added functions from libsystemd-journal. The interface was
2387 changed to bring the low level interface in s.j._Reader
2388 closer to the C API, and the high level interface in
2389 s.j.Reader was updated to wrap and convert all data about
2392 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
2393 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
2394 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
2395 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
2396 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
2397 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2401 * journalctl --update-catalog now understands a new --root=
2402 option to operate on catalogs found in a different root
2405 * During shutdown after systemd has terminated all running
2406 services a final killing loop kills all remaining left-over
2407 processes. We will now print the name of these processes
2408 when we send SIGKILL to them, since this usually indicates a
2411 * If /etc/crypttab refers to password files stored on
2412 configured mount points automatic dependencies will now be
2413 generated to ensure the specific mount is established first
2414 before the key file is attempted to be read.
2416 * 'systemctl status' will now show information about the
2417 network sockets a socket unit is listening on.
2419 * 'systemctl status' will also shown information about any
2420 drop-in configuration file for units. (Drop-In configuration
2421 files in this context are files such as
2422 /etc/systemd/systemd/foobar.service.d/*.conf)
2424 * systemd-cgtop now optionally shows summed up CPU times of
2425 cgroups. Press '%' while running cgtop to switch between
2426 percentage and absolute mode. This is useful to determine
2427 which cgroups use up the most CPU time over the entire
2428 runtime of the system. systemd-cgtop has also been updated
2429 to be 'pipeable' for processing with further shell tools.
2431 * 'hostnamectl set-hostname' will now allow setting of FQDN
2434 * The formatting and parsing of time span values has been
2435 changed. The parser now understands fractional expressions
2436 such as "5.5h". The formatter will now output fractional
2437 expressions for all time spans under 1min, i.e. "5.123456s"
2438 rather than "5s 123ms 456us". For time spans under 1s
2439 millisecond values are shown, for those under 1ms
2440 microsecond values are shown. This should greatly improve
2441 all time-related output of systemd.
2443 * libsystemd-login and libsystemd-journal gained new
2444 functions for querying the poll() events mask and poll()
2445 timeout value for integration into arbitrary event
2448 * localectl gained the ability to list available X11 keymaps
2449 (models, layouts, variants, options).
2451 * 'systemd-analyze dot' gained the ability to filter for
2452 specific units via shell-style globs, to create smaller,
2453 more useful graphs. I.e. it is now possible to create simple
2454 graphs of all the dependencies between only target units, or
2455 of all units that Avahi has dependencies with.
2457 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2458 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2459 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2460 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2461 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2462 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2463 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2467 * The boot-time readahead implementation for rotating media
2468 will now read the read-ahead data in multiple passes which
2469 consist of all read requests made in equidistant time
2470 intervals. This means instead of strictly reading read-ahead
2471 data in its physical order on disk we now try to find a
2472 middle ground between physical and access time order.
2474 * /etc/os-release files gained a new BUILD_ID= field for usage
2475 on operating systems that provide continuous builds of OS
2478 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
2479 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
2480 William Douglas, Zbigniew Jędrzejewski-Szmek
2484 * systemd-python gained an API exposing libsystemd-daemon.
2486 * The SMACK setup logic gained support for uploading CIPSO
2489 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
2490 ReadOnlyDirectories= and InaccessibleDirectories= has
2491 changed. The private /tmp and /var/tmp directories are now
2492 shared by all processes of a service (which means
2493 ExecStartPre= may now leave data in /tmp that ExecStart= of
2494 the same service can still access). When a service is
2495 stopped its temporary directories are immediately deleted
2496 (normal clean-up with tmpfiles is still done in addition to
2499 * By default, systemd will now set a couple of sysctl
2500 variables in the kernel: the safe sysrq options are turned
2501 on, IP route verification is turned on, and source routing
2502 disabled. The recently added hardlink and softlink
2503 protection of the kernel is turned on. These settings should
2504 be reasonably safe, and good defaults for all new systems.
2506 * The predictable network naming logic may now be turned off
2507 with a new kernel command line switch: net.ifnames=0.
2509 * A new libsystemd-bus module has been added that implements a
2510 pretty complete D-Bus client library. For details see:
2512 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
2514 * journald will now explicitly flush the journal files to disk
2515 at the latest 5min after each write. The file will then also
2516 be marked offline until the next write. This should increase
2517 reliability in case of a crash. The synchronization delay
2518 can be configured via SyncIntervalSec= in journald.conf.
2520 * There's a new remote-fs-setup.target unit that can be used
2521 to pull in specific services when at least one remote file
2522 system is to be mounted.
2524 * There are new targets timers.target and paths.target as
2525 canonical targets to pull user timer and path units in
2526 from. This complements sockets.target with a similar
2527 purpose for socket units.
2529 * libudev gained a new call udev_device_set_attribute_value()
2530 to set sysfs attributes of a device.
2532 * The udev daemon now sets the default number of worker
2533 processes executed in parallel based on the number of available
2534 CPUs instead of the amount of available RAM. This is supposed
2535 to provide a more reliable default and limit a too aggressive
2536 paralellism for setups with 1000s of devices connected.
2538 Contributions from: Auke Kok, Colin Walters, Cristian
2539 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
2540 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
2541 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
2542 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
2543 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
2544 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
2545 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
2546 Zbigniew Jędrzejewski-Szmek
2550 * Configuration of unit files may now be extended via drop-in
2551 files without having to edit/override the unit files
2552 themselves. More specifically, if the administrator wants to
2553 change one value for a service file foobar.service he can
2554 now do so by dropping in a configuration snippet into
2555 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
2556 will load all these snippets and apply them on top of the
2557 main unit configuration file, possibly extending or
2558 overriding its settings. Using these drop-in snippets is
2559 generally nicer than the two earlier options for changing
2560 unit files locally: copying the files from
2561 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
2562 them there; or creating a new file in /etc/systemd/system/
2563 that incorporates the original one via ".include". Drop-in
2564 snippets into these .d/ directories can be placed in any
2565 directory systemd looks for units in, and the usual
2566 overriding semantics between /usr/lib, /etc and /run apply
2569 * Most unit file settings which take lists of items can now be
2570 reset by assigning the empty string to them. For example,
2571 normally, settings such as Environment=FOO=BAR append a new
2572 environment variable assignment to the environment block,
2573 each time they are used. By assigning Environment= the empty
2574 string the environment block can be reset to empty. This is
2575 particularly useful with the .d/*.conf drop-in snippets
2576 mentioned above, since this adds the ability to reset list
2577 settings from vendor unit files via these drop-ins.
2579 * systemctl gained a new "list-dependencies" command for
2580 listing the dependencies of a unit recursively.
2582 * Inhibitors are now honored and listed by "systemctl
2583 suspend", "systemctl poweroff" (and similar) too, not only
2584 GNOME. These commands will also list active sessions by
2587 * Resource limits (as exposed by the various control group
2588 controllers) can now be controlled dynamically at runtime
2589 for all units. More specifically, you can now use a command
2590 like "systemctl set-cgroup-attr foobar.service cpu.shares
2591 2000" to alter the CPU shares a specific service gets. These
2592 settings are stored persistently on disk, and thus allow the
2593 administrator to easily adjust the resource usage of
2594 services with a few simple commands. This dynamic resource
2595 management logic is also available to other programs via the
2596 bus. Almost any kernel cgroup attribute and controller is
2599 * systemd-vconsole-setup will now copy all font settings to
2600 all allocated VTs, where it previously applied them only to
2603 * libsystemd-login gained the new sd_session_get_tty() API
2606 * This release drops support for a few legacy or
2607 distribution-specific LSB facility names when parsing init
2608 scripts: $x-display-manager, $mail-transfer-agent,
2609 $mail-transport-agent, $mail-transfer-agent, $smtp,
2610 $null. Also, the mail-transfer-agent.target unit backing
2611 this has been removed. Distributions which want to retain
2612 compatibility with this should carry the burden for
2613 supporting this themselves and patch support for these back
2614 in, if they really need to. Also, the facilities $syslog and
2615 $local_fs are now ignored, since systemd does not support
2616 early-boot LSB init scripts anymore, and these facilities
2617 are implied anyway for normal services. syslog.target has
2620 * There are new bus calls on PID1's Manager object for
2621 cancelling jobs, and removing snapshot units. Previously,
2622 both calls were only available on the Job and Snapshot
2625 * systemd-journal-gatewayd gained SSL support.
2627 * The various "environment" files, such as /etc/locale.conf
2628 now support continuation lines with a backslash ("\") as
2629 last character in the line, similarly in style (but different)
2630 to how this is supported in shells.
2632 * For normal user processes the _SYSTEMD_USER_UNIT= field is
2633 now implicitly appended to every log entry logged. systemctl
2634 has been updated to filter by this field when operating on a
2635 user systemd instance.
2637 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
2638 CAP_AUDIT_CONTROL capabilities to the capabilities set for
2639 the container. This makes it easier to boot unmodified
2640 Fedora systems in a container, which however still requires
2641 audit=0 to be passed on the kernel command line. Auditing in
2642 kernel and userspace is unfortunately still too broken in
2643 context of containers, hence we recommend compiling it out
2644 of the kernel or using audit=0. Hopefully this will be fixed
2645 one day for good in the kernel.
2647 * nspawn gained the new --bind= and --bind-ro= parameters to
2648 bind mount specific directories from the host into the
2651 * nspawn will now mount its own devpts file system instance
2652 into the container, in order not to leak pty devices from
2653 the host into the container.
2655 * systemd will now read the firmware boot time performance
2656 information from the EFI variables, if the used boot loader
2657 supports this, and takes it into account for boot performance
2658 analysis via "systemd-analyze". This is currently supported
2659 only in conjunction with Gummiboot, but could be supported
2660 by other boot loaders too. For details see:
2662 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
2664 * A new generator has been added that automatically mounts the
2665 EFI System Partition (ESP) to /boot, if that directory
2666 exists, is empty, and no other file system has been
2667 configured to be mounted there.
2669 * logind will now send out PrepareForSleep(false) out
2670 unconditionally, after coming back from suspend. This may be
2671 used by applications as asynchronous notification for
2672 system resume events.
2674 * "systemctl unlock-sessions" has been added, that allows
2675 unlocking the screens of all user sessions at once, similar
2676 to how "systemctl lock-sessions" already locked all users
2677 sessions. This is backed by a new D-Bus call UnlockSessions().
2679 * "loginctl seat-status" will now show the master device of a
2680 seat. (i.e. the device of a seat that needs to be around for
2681 the seat to be considered available, usually the graphics
2684 * tmpfiles gained a new "X" line type, that allows
2685 configuration of files and directories (with wildcards) that
2686 shall be excluded from automatic cleanup ("aging").
2688 * udev default rules set the device node permissions now only
2689 at "add" events, and do not change them any longer with a
2690 later "change" event.
2692 * The log messages for lid events and power/sleep keypresses
2693 now carry a message ID.
2695 * We now have a substantially larger unit test suite, but this
2696 continues to be work in progress.
2698 * udevadm hwdb gained a new --root= parameter to change the
2699 root directory to operate relative to.
2701 * logind will now issue a background sync() request to the kernel
2702 early at shutdown, so that dirty buffers are flushed to disk early
2703 instead of at the last moment, in order to optimize shutdown
2706 * A new bootctl tool has been added that is an interface for
2707 certain boot loader operations. This is currently a preview
2708 and is likely to be extended into a small mechanism daemon
2709 like timedated, localed, hostnamed, and can be used by
2710 graphical UIs to enumerate available boot options, and
2711 request boot into firmware operations.
2713 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2714 the rest of the package. It also has been updated to work
2715 correctly in initrds.
2717 * Policykit previously has been runtime optional, and is now
2718 also compile time optional via a configure switch.
2720 * systemd-analyze has been reimplemented in C. Also "systemctl
2721 dot" has moved into systemd-analyze.
2723 * "systemctl status" with no further parameters will now print
2724 the status of all active or failed units.
2726 * Operations such as "systemctl start" can now be executed
2727 with a new mode "--irreversible" which may be used to queue
2728 operations that cannot accidentally be reversed by a later
2729 job queuing. This is by default used to make shutdown
2730 requests more robust.
2732 * The Python API of systemd now gained a new module for
2733 reading journal files.
2735 * A new tool kernel-install has been added that can install
2736 kernel images according to the Boot Loader Specification:
2738 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2740 * Boot time console output has been improved to provide
2741 animated boot time output for hanging jobs.
2743 * A new tool systemd-activate has been added which can be used
2744 to test socket activation with, directly from the command
2745 line. This should make it much easier to test and debug
2746 socket activation in daemons.
2748 * journalctl gained a new "--reverse" (or -r) option to show
2749 journal output in reverse order (i.e. newest line first).
2751 * journalctl gained a new "--pager-end" (or -e) option to jump
2752 to immediately jump to the end of the journal in the
2753 pager. This is only supported in conjunction with "less".
2755 * journalctl gained a new "--user-unit=" option, that works
2756 similarly to "--unit=" but filters for user units rather than
2759 * A number of unit files to ease adoption of systemd in
2760 initrds has been added. This moves some minimal logic from
2761 the various initrd implementations into systemd proper.
2763 * The journal files are now owned by a new group
2764 "systemd-journal", which exists specifically to allow access
2765 to the journal, and nothing else. Previously, we used the
2766 "adm" group for that, which however possibly covers more
2767 than just journal/log file access. This new group is now
2768 already used by systemd-journal-gatewayd to ensure this
2769 daemon gets access to the journal files and as little else
2770 as possible. Note that "make install" will also set FS ACLs
2771 up for /var/log/journal to give "adm" and "wheel" read
2772 access to it, in addition to "systemd-journal" which owns
2773 the journal files. We recommend that packaging scripts also
2774 add read access to "adm" + "wheel" to /var/log/journal, and
2775 all existing/future journal files. To normal users and
2776 administrators little changes, however packagers need to
2777 ensure to create the "systemd-journal" system group at
2778 package installation time.
2780 * The systemd-journal-gatewayd now runs as unprivileged user
2781 systemd-journal-gateway:systemd-journal-gateway. Packaging
2782 scripts need to create these system user/group at
2785 * timedated now exposes a new boolean property CanNTP that
2786 indicates whether a local NTP service is available or not.
2788 * systemd-detect-virt will now also detect xen PVs
2790 * The pstore file system is now mounted by default, if it is
2793 * In addition to the SELinux and IMA policies we will now also
2794 load SMACK policies at early boot.
2796 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2797 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2798 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2799 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2800 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2801 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2802 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2803 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2804 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2805 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2806 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2807 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2808 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2809 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2813 * Timer units now support calendar time events in addition to
2814 monotonic time events. That means you can now trigger a unit
2815 based on a calendar time specification such as "Thu,Fri
2816 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2817 or fifth day of any month of the year 2013, given that it is
2818 a thursday or friday. This brings timer event support
2819 considerably closer to cron's capabilities. For details on
2820 the supported calendar time specification language see
2823 * udev now supports a number of different naming policies for
2824 network interfaces for predictable names, and a combination
2825 of these policies is now the default. Please see this wiki
2826 document for details:
2828 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2830 * Auke Kok's bootchart implementation has been added to the
2831 systemd tree. It is an optional component that can graph the
2832 boot in quite some detail. It is one of the best bootchart
2833 implementations around and minimal in its code and
2836 * nss-myhostname has been integrated into the systemd source
2837 tree. nss-myhostname guarantees that the local hostname
2838 always stays resolvable via NSS. It has been a weak
2839 requirement of systemd-hostnamed since a long time, and
2840 since its code is actually trivial we decided to just
2841 include it in systemd's source tree. It can be turned off
2842 with a configure switch.
2844 * The read-ahead logic is now capable of properly detecting
2845 whether a btrfs file system is on SSD or rotating media, in
2846 order to optimize the read-ahead scheme. Previously, it was
2847 only capable of detecting this on traditional file systems
2850 * In udev, additional device properties are now read from the
2851 IAB in addition to the OUI database. Also, Bluetooth company
2852 identities are attached to the devices as well.
2854 * In service files %U may be used as specifier that is
2855 replaced by the configured user name of the service.
2857 * nspawn may now be invoked without a controlling TTY. This
2858 makes it suitable for invocation as its own service. This
2859 may be used to set up a simple containerized server system
2860 using only core OS tools.
2862 * systemd and nspawn can now accept socket file descriptors
2863 when they are started for socket activation. This enables
2864 implementation of socket activated nspawn
2865 containers. i.e. think about autospawning an entire OS image
2866 when the first SSH or HTTP connection is received. We expect
2867 that similar functionality will also be added to libvirt-lxc
2870 * journalctl will now suppress ANSI color codes when
2871 presenting log data.
2873 * systemctl will no longer show control group information for
2874 a unit if a the control group is empty anyway.
2876 * logind can now automatically suspend/hibernate/shutdown the
2879 * /etc/machine-info and hostnamed now also expose the chassis
2880 type of the system. This can be used to determine whether
2881 the local system is a laptop, desktop, handset or
2882 tablet. This information may either be configured by the
2883 user/vendor or is automatically determined from ACPI and DMI
2884 information if possible.
2886 * A number of PolicyKit actions are now bound together with
2887 "imply" rules. This should simplify creating UIs because
2888 many actions will now authenticate similar ones as well.
2890 * Unit files learnt a new condition ConditionACPower= which
2891 may be used to conditionalize a unit depending on whether an
2892 AC power source is connected or not, of whether the system
2893 is running on battery power.
2895 * systemctl gained a new "is-failed" verb that may be used in
2896 shell scripts and suchlike to check whether a specific unit
2897 is in the "failed" state.
2899 * The EnvironmentFile= setting in unit files now supports file
2900 globbing, and can hence be used to easily read a number of
2901 environment files at once.
2903 * systemd will no longer detect and recognize specific
2904 distributions. All distribution-specific #ifdeffery has been
2905 removed, systemd is now fully generic and
2906 distribution-agnostic. Effectively, not too much is lost as
2907 a lot of the code is still accessible via explicit configure
2908 switches. However, support for some distribution specific
2909 legacy configuration file formats has been dropped. We
2910 recommend distributions to simply adopt the configuration
2911 files everybody else uses now and convert the old
2912 configuration from packaging scripts. Most distributions
2913 already did that. If that's not possible or desirable,
2914 distributions are welcome to forward port the specific
2915 pieces of code locally from the git history.
2917 * When logging a message about a unit systemd will now always
2918 log the unit name in the message meta data.
2920 * localectl will now also discover system locale data that is
2921 not stored in locale archives, but directly unpacked.
2923 * logind will no longer unconditionally use framebuffer
2924 devices as seat masters, i.e. as devices that are required
2925 to be existing before a seat is considered preset. Instead,
2926 it will now look for all devices that are tagged as
2927 "seat-master" in udev. By default framebuffer devices will
2928 be marked as such, but depending on local systems other
2929 devices might be marked as well. This may be used to
2930 integrate graphics cards using closed source drivers (such
2931 as NVidia ones) more nicely into logind. Note however, that
2932 we recommend using the open source NVidia drivers instead,
2933 and no udev rules for the closed-source drivers will be
2934 shipped from us upstream.
2936 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2937 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2938 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2939 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2940 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2941 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2942 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2943 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2944 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2945 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2946 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2947 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2952 * udev gained support for loading additional device properties
2953 from an indexed database that is keyed by vendor/product IDs
2954 and similar device identifiers. For the beginning this
2955 "hwdb" is populated with data from the well-known PCI and
2956 USB database, but also includes PNP, ACPI and OID data. In
2957 the longer run this indexed database shall grow into
2958 becoming the one central database for non-essential
2959 userspace device metadata. Previously, data from the PCI/USB
2960 database was only attached to select devices, since the
2961 lookup was a relatively expensive operation due to O(n) time
2962 complexity (with n being the number of entries in the
2963 database). Since this is now O(1), we decided to add in this
2964 data for all devices where this is available, by
2965 default. Note that the indexed database needs to be rebuilt
2966 when new data files are installed. To achieve this you need
2967 to update your packaging scripts to invoke "udevadm hwdb
2968 --update" after installation of hwdb data files. For
2969 RPM-based distributions we introduced the new
2970 %udev_hwdb_update macro for this purpose.
2972 * The Journal gained support for the "Message Catalog", an
2973 indexed database to link up additional information with
2974 journal entries. For further details please check:
2976 http://www.freedesktop.org/wiki/Software/systemd/catalog
2978 The indexed message catalog database also needs to be
2979 rebuilt after installation of message catalog files. Use
2980 "journalctl --update-catalog" for this. For RPM-based
2981 distributions we introduced the %journal_catalog_update
2982 macro for this purpose.
2984 * The Python Journal bindings gained support for the standard
2985 Python logging framework.
2987 * The Journal API gained new functions for checking whether
2988 the underlying file system of a journal file is capable of
2989 properly reporting file change notifications, or whether
2990 applications that want to reflect journal changes "live"
2991 need to recheck journal files continuously in appropriate
2994 * It is now possible to set the "age" field for tmpfiles
2995 entries to 0, indicating that files matching this entry
2996 shall always be removed when the directories are cleaned up.
2998 * coredumpctl gained a new "gdb" verb which invokes gdb
2999 right-away on the selected coredump.
3001 * There's now support for "hybrid sleep" on kernels that
3002 support this, in addition to "suspend" and "hibernate". Use
3003 "systemctl hybrid-sleep" to make use of this.
3005 * logind's HandleSuspendKey= setting (and related settings)
3006 now gained support for a new "lock" setting to simply
3007 request the screen lock on all local sessions, instead of
3008 actually executing a suspend or hibernation.
3010 * systemd will now mount the EFI variables file system by
3013 * Socket units now gained support for configuration of the
3014 SMACK security label.
3016 * timedatectl will now output the time of the last and next
3017 daylight saving change.
3019 * We dropped support for various legacy and distro-specific
3020 concepts, such as insserv, early-boot SysV services
3021 (i.e. those for non-standard runlevels such as 'b' or 'S')
3022 or ArchLinux /etc/rc.conf support. We recommend the
3023 distributions who still need support this to either continue
3024 to maintain the necessary patches downstream, or find a
3025 different solution. (Talk to us if you have questions!)
3027 * Various systemd components will now bypass PolicyKit checks
3028 for root and otherwise handle properly if PolicyKit is not
3029 found to be around. This should fix most issues for
3030 PolicyKit-less systems. Quite frankly this should have been
3031 this way since day one. It is absolutely our intention to
3032 make systemd work fine on PolicyKit-less systems, and we
3033 consider it a bug if something does not work as it should if
3034 PolicyKit is not around.
3036 * For embedded systems it is now possible to build udev and
3037 systemd without blkid and/or kmod support.
3039 * "systemctl switch-root" is now capable of switching root
3040 more than once. I.e. in addition to transitions from the
3041 initrd to the host OS it is now possible to transition to
3042 further OS images from the host. This is useful to implement
3043 offline updating tools.
3045 * Various other additions have been made to the RPM macros
3046 shipped with systemd. Use %udev_rules_update() after
3047 installing new udev rules files. %_udevhwdbdir,
3048 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
3049 %_sysctldir are now available which resolve to the right
3050 directories for packages to place various data files in.
3052 * journalctl gained the new --full switch (in addition to
3053 --all, to disable ellipsation for long messages.
3055 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
3056 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
3057 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
3058 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
3059 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
3060 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
3061 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
3062 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
3063 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
3067 * journalctl gained new --since= and --until= switches to
3068 filter by time. It also now supports nice filtering for
3069 units via --unit=/-u.
3071 * Type=oneshot services may use ExecReload= and do the
3074 * The journal daemon now supports time-based rotation and
3075 vacuuming, in addition to the usual disk-space based
3078 * The journal will now index the available field values for
3079 each field name. This enables clients to show pretty drop
3080 downs of available match values when filtering. The bash
3081 completion of journalctl has been updated
3082 accordingly. journalctl gained a new switch -F to list all
3083 values a certain field takes in the journal database.
3085 * More service events are now written as structured messages
3086 to the journal, and made recognizable via message IDs.
3088 * The timedated, localed and hostnamed mini-services which
3089 previously only provided support for changing time, locale
3090 and hostname settings from graphical DEs such as GNOME now
3091 also have a minimal (but very useful) text-based client
3092 utility each. This is probably the nicest way to changing
3093 these settings from the command line now, especially since
3094 it lists available options and is fully integrated with bash
3097 * There's now a new tool "systemd-coredumpctl" to list and
3098 extract coredumps from the journal.
3100 * We now install a README each in /var/log/ and
3101 /etc/rc.d/init.d explaining where the system logs and init
3102 scripts went. This hopefully should help folks who go to
3103 that dirs and look into the otherwise now empty void and
3104 scratch their heads.
3106 * When user-services are invoked (by systemd --user) the
3107 $MANAGERPID env var is set to the PID of systemd.
3109 * SIGRTMIN+24 when sent to a --user instance will now result
3110 in immediate termination of systemd.
3112 * gatewayd received numerous feature additions such as a
3113 "follow" mode, for live syncing and filtering.
3115 * browse.html now allows filtering and showing detailed
3116 information on specific entries. Keyboard navigation and
3117 mouse screen support has been added.
3119 * gatewayd/journalctl now supports HTML5/JSON
3120 Server-Sent-Events as output.
3122 * The SysV init script compatibility logic will now
3123 heuristically determine whether a script supports the
3124 "reload" verb, and only then make this available as
3127 * "systemctl status --follow" has been removed, use "journalctl
3130 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3131 have been removed since they are hardly useful to be
3134 * And I'd like to take the opportunity to specifically mention
3135 Zbigniew for his great contributions. Zbigniew, you rock!
3137 Contributions from: Andrew Eikum, Christian Hesse, Colin
3138 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3139 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3140 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3141 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3142 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3143 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3147 * If /etc/vconsole.conf is non-existent or empty we will no
3148 longer load any console font or key map at boot by
3149 default. Instead the kernel defaults will be left
3150 intact. This is definitely the right thing to do, as no
3151 configuration should mean no configuration, and hard-coding
3152 font names that are different on all archs is probably a bad
3153 idea. Also, the kernel default key map and font should be
3154 good enough for most cases anyway, and mostly identical to
3155 the userspace fonts/key maps we previously overloaded them
3156 with. If distributions want to continue to default to a
3157 non-kernel font or key map they should ship a default
3158 /etc/vconsole.conf with the appropriate contents.
3160 Contributions from: Colin Walters, Daniel J Walsh, Dave
3161 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3162 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3166 * journalctl gained a new --cursor= switch to show entries
3167 starting from the specified location in the journal.
3169 * We now enforce a size limit on journal entry fields exported
3170 with "-o json" in journalctl. Fields larger than 4K will be
3171 assigned null. This can be turned off with --all.
3173 * An (optional) journal gateway daemon is now available as
3174 "systemd-journal-gatewayd.service". This service provides
3175 access to the journal via HTTP and JSON. This functionality
3176 will be used to implement live log synchronization in both
3177 pull and push modes, but has various other users too, such
3178 as easy log access for debugging of embedded devices. Right
3179 now it is already useful to retrieve the journal via HTTP:
3181 # systemctl start systemd-journal-gatewayd.service
3182 # wget http://localhost:19531/entries
3184 This will download the journal contents in a
3185 /var/log/messages compatible format. The same as JSON:
3187 # curl -H"Accept: application/json" http://localhost:19531/entries
3189 This service is also accessible via a web browser where a
3190 single static HTML5 app is served that uses the JSON logic
3191 to enable the user to do some basic browsing of the
3192 journal. This will be extended later on. Here's an example
3193 screenshot of this app in its current state:
3195 http://0pointer.de/public/journal-gatewayd
3197 Contributions from: Kay Sievers, Lennart Poettering, Robert
3198 Milasan, Tom Gundersen
3202 * The bash completion logic is now available for journalctl
3205 * We do not mount the "cpuset" controller anymore together with
3206 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
3207 started if no parameters are assigned to it. "cpuset" hence
3208 broke code that assumed it it could create "cpu" groups and
3211 * journalctl -f will now subscribe to terminal size changes,
3212 and line break accordingly.
3214 Contributions from: Dave Reisner, Kay Sievers, Lennart
3215 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3219 * nspawn will now create a symlink /etc/localtime in the
3220 container environment, copying the host's timezone
3221 setting. Previously this has been done via a bind mount, but
3222 since symlinks cannot be bind mounted this has now been
3223 changed to create/update the appropriate symlink.
3225 * journalctl -n's line number argument is now optional, and
3226 will default to 10 if omitted.
3228 * journald will now log the maximum size the journal files may
3229 take up on disk. This is particularly useful if the default
3230 built-in logic of determining this parameter from the file
3231 system size is used. Use "systemctl status
3232 systemd-journald.service" to see this information.
3234 * The multi-seat X wrapper tool has been stripped down. As X
3235 is now capable of enumerating graphics devices via udev in a
3236 seat-aware way the wrapper is not strictly necessary
3237 anymore. A stripped down temporary stop-gap is still shipped
3238 until the upstream display managers have been updated to
3239 fully support the new X logic. Expect this wrapper to be
3240 removed entirely in one of the next releases.
3242 * HandleSleepKey= in logind.conf has been split up into
3243 HandleSuspendKey= and HandleHibernateKey=. The old setting
3244 is not available anymore. X11 and the kernel are
3245 distuingishing between these keys and we should too. This
3246 also means the inhibition lock for these keys has been split
3249 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
3250 Poettering, Lukas Nykryn, Václav Pavlín
3254 * Whenever a unit changes state we will now log this to the
3255 journal and show along the unit's own log output in
3258 * ConditionPathIsMountPoint= can now properly detect bind
3259 mount points too. (Previously, a bind mount of one file
3260 system to another place in the same file system could not be
3261 detected as mount, since they shared struct stat's st_dev
3264 * We will now mount the cgroup controllers cpu, cpuacct,
3265 cpuset and the controllers net_cls, net_prio together by
3268 * nspawn containers will now have a virtualized boot
3269 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
3270 over with a randomized ID at container initialization). This
3271 has the effect of making "journalctl -b" do the right thing
3274 * The JSON output journal serialization has been updated not
3275 to generate "endless" list objects anymore, but rather one
3276 JSON object per line. This is more in line how most JSON
3277 parsers expect JSON objects. The new output mode
3278 "json-pretty" has been added to provide similar output, but
3279 neatly aligned for readability by humans.
3281 * We dropped all explicit sync() invocations in the shutdown
3282 code. The kernel does this implicitly anyway in the kernel
3283 reboot() syscall. halt(8)'s -n option is now a compatibility
3286 * We now support virtualized reboot() in containers, as
3287 supported by newer kernels. We will fall back to exit() if
3288 CAP_SYS_REBOOT is not available to the container. Also,
3289 nspawn makes use of this now and will actually reboot the
3290 container if the containerized OS asks for that.
3292 * journalctl will only show local log output by default
3293 now. Use --merge (-m) to show remote log output, too.
3295 * libsystemd-journal gained the new sd_journal_get_usage()
3296 call to determine the current disk usage of all journal
3297 files. This is exposed in the new "journalctl --disk-usage"
3300 * journald gained a new configuration setting SplitMode= in
3301 journald.conf which may be used to control how user journals
3302 are split off. See journald.conf(5) for details.
3304 * A new condition type ConditionFileNotEmpty= has been added.
3306 * tmpfiles' "w" lines now support file globbing, to write
3307 multiple files at once.
3309 * We added Python bindings for the journal submission
3310 APIs. More Python APIs for a number of selected APIs will
3311 likely follow. Note that we intend to add native bindings
3312 only for the Python language, as we consider it common
3313 enough to deserve bindings shipped within systemd. There are
3314 various projects outside of systemd that provide bindings
3315 for languages such as PHP or Lua.
3317 * Many conditions will now resolve specifiers such as %i. In
3318 addition, PathChanged= and related directives of .path units
3319 now support specifiers as well.
3321 * There's now a new RPM macro definition for the system preset
3324 * journald will now warn if it ca not forward a message to the
3325 syslog daemon because its socket is full.
3327 * timedated will no longer write or process /etc/timezone,
3328 except on Debian. As we do not support late mounted /usr
3329 anymore /etc/localtime always being a symlink is now safe,
3330 and hence the information in /etc/timezone is not necessary
3333 * logind will now always reserve one VT for a text getty (VT6
3334 by default). Previously if more than 6 X sessions where
3335 started they took up all the VTs with auto-spawned gettys,
3336 so that no text gettys were available anymore.
3338 * udev will now automatically inform the btrfs kernel logic
3339 about btrfs RAID components showing up. This should make
3340 simple hotplug based btrfs RAID assembly work.
3342 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
3343 (but not for its children which will stay at the kernel
3344 default). This should allow setups with a lot more listening
3347 * systemd will now always pass the configured timezone to the
3348 kernel at boot. timedated will do the same when the timezone
3351 * logind's inhibition logic has been updated. By default,
3352 logind will now handle the lid switch, the power and sleep
3353 keys all the time, even in graphical sessions. If DEs want
3354 to handle these events on their own they should take the new
3355 handle-power-key, handle-sleep-key and handle-lid-switch
3356 inhibitors during their runtime. A simple way to achiveve
3357 that is to invoke the DE wrapped in an invocation of:
3359 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
3361 * Access to unit operations is now checked via SELinux taking
3362 the unit file label and client process label into account.
3364 * systemd will now notify the administrator in the journal
3365 when he over-mounts a non-empty directory.
3367 * There are new specifiers that are resolved in unit files,
3368 for the host name (%H), the machine ID (%m) and the boot ID
3371 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
3372 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
3373 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
3374 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
3375 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
3376 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3377 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
3381 * Support for reading structured kernel messages from
3382 /dev/kmsg has now been added and is enabled by default.
3384 * Support for reading kernel messages from /proc/kmsg has now
3385 been removed. If you want kernel messages in the journal
3386 make sure to run a recent kernel (>= 3.5) that supports
3387 reading structured messages from /dev/kmsg (see
3388 above). /proc/kmsg is now exclusive property of classic
3389 syslog daemons again.
3391 * The libudev API gained the new
3392 udev_device_new_from_device_id() call.
3394 * The logic for file system namespace (ReadOnlyDirectory=,
3395 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
3396 require pivot_root() anymore. This means fewer temporary
3397 directories are created below /tmp for this feature.
3399 * nspawn containers will now see and receive all submounts
3400 made on the host OS below the root file system of the
3403 * Forward Secure Sealing is now supported for Journal files,
3404 which provide cryptographical sealing of journal files so
3405 that attackers cannot alter log history anymore without this
3406 being detectable. Lennart will soon post a blog story about
3407 this explaining it in more detail.
3409 * There are two new service settings RestartPreventExitStatus=
3410 and SuccessExitStatus= which allow configuration of exit
3411 status (exit code or signal) which will be excepted from the
3412 restart logic, resp. consider successful.
3414 * journalctl gained the new --verify switch that can be used
3415 to check the integrity of the structure of journal files and
3416 (if Forward Secure Sealing is enabled) the contents of
3419 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3420 and similar symlinks pre-created. This makes running shells
3421 as container init process a lot more fun.
3423 * The fstab support can now handle PARTUUID= and PARTLABEL=
3426 * A new ConditionHost= condition has been added to match
3427 against the hostname (with globs) and machine ID. This is
3428 useful for clusters where a single OS image is used to
3429 provision a large number of hosts which shall run slightly
3430 different sets of services.
3432 * Services which hit the restart limit will now be placed in a
3435 Contributions from: Bertram Poettering, Dave Reisner, Huang
3436 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3437 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3441 * When running in --user mode systemd will now become a
3442 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3443 tree a lot more organized.
3445 * A new PartOf= unit dependency type has been introduced that
3446 may be used to group services in a natural way.
3448 * "systemctl enable" may now be used to enable instances of
3451 * journalctl now prints error log levels in red, and
3452 warning/notice log levels in bright white. It also supports
3453 filtering by log level now.
3455 * cgtop gained a new -n switch (similar to top), to configure
3456 the maximum number of iterations to run for. It also gained
3457 -b, to run in batch mode (accepting no input).
3459 * The suffix ".service" may now be omitted on most systemctl
3460 command lines involving service unit names.
3462 * There's a new bus call in logind to lock all sessions, as
3463 well as a loginctl verb for it "lock-sessions".
3465 * libsystemd-logind.so gained a new call sd_journal_perror()
3466 that works similar to libc perror() but logs to the journal
3467 and encodes structured information about the error number.
3469 * /etc/crypttab entries now understand the new keyfile-size=
3472 * shutdown(8) now can send a (configurable) wall message when
3473 a shutdown is cancelled.
3475 * The mount propagation mode for the root file system will now
3476 default to "shared", which is useful to make containers work
3477 nicely out-of-the-box so that they receive new mounts from
3478 the host. This can be undone locally by running "mount
3479 --make-rprivate /" if needed.
3481 * The prefdm.service file has been removed. Distributions
3482 should maintain this unit downstream if they intend to keep
3483 it around. However, we recommend writing normal unit files
3484 for display managers instead.
3486 * Since systemd is a crucial part of the OS we will now
3487 default to a number of compiler switches that improve
3488 security (hardening) such as read-only relocations, stack
3489 protection, and suchlike.
3491 * The TimeoutSec= setting for services is now split into
3492 TimeoutStartSec= and TimeoutStopSec= to allow configuration
3493 of individual time outs for the start and the stop phase of
3496 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
3497 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
3498 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
3499 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
3500 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
3501 Gundersen, Zbigniew Jędrzejewski-Szmek
3505 * The journal and id128 C APIs are now fully documented as man
3508 * Extra safety checks have been added when transitioning from
3509 the initial RAM disk to the main system to avoid accidental
3512 * /etc/crypttab entries now understand the new keyfile-offset=
3515 * systemctl -t can now be used to filter by unit load state.
3517 * The journal C API gained the new sd_journal_wait() call to
3518 make writing synchronous journal clients easier.
3520 * journalctl gained the new -D switch to show journals from a
3523 * journalctl now displays a special marker between log
3524 messages of two different boots.
3526 * The journal is now explicitly flushed to /var via a service
3527 systemd-journal-flush.service, rather than implicitly simply
3528 by seeing /var/log/journal to be writable.
3530 * journalctl (and the journal C APIs) can now match for much
3531 more complex expressions, with alternatives and
3534 * When transitioning from the initial RAM disk to the main
3535 system we will now kill all processes in a killing spree to
3536 ensure no processes stay around by accident.
3538 * Three new specifiers may be used in unit files: %u, %h, %s
3539 resolve to the user name, user home directory resp. user
3540 shell. This is useful for running systemd user instances.
3542 * We now automatically rotate journal files if their data
3543 object hash table gets a fill level > 75%. We also size the
3544 hash table based on the configured maximum file size. This
3545 together should lower hash collisions drastically and thus
3546 speed things up a bit.
3548 * journalctl gained the new "--header" switch to introspect
3549 header data of journal files.
3551 * A new setting SystemCallFilters= has been added to services
3552 which may be used to apply blacklists or whitelists to
3553 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
3555 * nspawn gained a new --link-journal= switch (and quicker: -j)
3556 to link the container journal with the host. This makes it
3557 very easy to centralize log viewing on the host for all
3558 guests while still keeping the journal files separated.
3560 * Many bugfixes and optimizations
3562 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
3563 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
3564 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
3569 * Several tools now understand kernel command line arguments,
3570 which are only read when run in an initial RAM disk. They
3571 usually follow closely their normal counterparts, but are
3574 * There's a new tool to analyze the readahead files that are
3575 automatically generated at boot. Use:
3577 /usr/lib/systemd/systemd-readahead analyze /.readahead
3579 * We now provide an early debug shell on tty9 if this enabled. Use:
3581 systemctl enable debug-shell.service
3583 * All plymouth related units have been moved into the Plymouth
3584 package. Please make sure to upgrade your Plymouth version
3587 * systemd-tmpfiles now supports getting passed the basename of
3588 a configuration file only, in which case it will look for it
3589 in all appropriate directories automatically.
3591 * udevadm info now takes a /dev or /sys path as argument, and
3592 does the right thing. Example:
3594 udevadm info /dev/sda
3595 udevadm info /sys/class/block/sda
3597 * systemctl now prints a warning if a unit is stopped but a
3598 unit that might trigger it continues to run. Example: a
3599 service is stopped but the socket that activates it is left
3602 * "systemctl status" will now mention if the log output was
3603 shortened due to rotation since a service has been started.
3605 * The journal API now exposes functions to determine the
3606 "cutoff" times due to rotation.
3608 * journald now understands SIGUSR1 and SIGUSR2 for triggering
3609 immediately flushing of runtime logs to /var if possible,
3610 resp. for triggering immediate rotation of the journal
3613 * It is now considered an error if a service is attempted to
3614 be stopped that is not loaded.
3616 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
3618 * systemd-analyze now supports Python 3
3620 * tmpfiles now supports cleaning up directories via aging
3621 where the first level dirs are always kept around but
3622 directories beneath it automatically aged. This is enabled
3623 by prefixing the age field with '~'.
3625 * Seat objects now expose CanGraphical, CanTTY properties
3626 which is required to deal with very fast bootups where the
3627 display manager might be running before the graphics drivers
3628 completed initialization.
3630 * Seat objects now expose a State property.
3632 * We now include RPM macros for service enabling/disabling
3633 based on the preset logic. We recommend RPM based
3634 distributions to make use of these macros if possible. This
3635 makes it simpler to reuse RPM spec files across
3638 * We now make sure that the collected systemd unit name is
3639 always valid when services log to the journal via
3642 * There's a new man page kernel-command-line(7) detailing all
3643 command line options we understand.
3645 * The fstab generator may now be disabled at boot by passing
3646 fstab=0 on the kernel command line.
3648 * A new kernel command line option modules-load= is now understood
3649 to load a specific kernel module statically, early at boot.
3651 * Unit names specified on the systemctl command line are now
3652 automatically escaped as needed. Also, if file system or
3653 device paths are specified they are automatically turned
3654 into the appropriate mount or device unit names. Example:
3656 systemctl status /home
3657 systemctl status /dev/sda
3659 * The SysVConsole= configuration option has been removed from
3660 system.conf parsing.
3662 * The SysV search path is no longer exported on the D-Bus
3665 * The Names= option is been removed from unit file parsing.
3667 * There's a new man page bootup(7) detailing the boot process.
3669 * Every unit and every generator we ship with systemd now
3670 comes with full documentation. The self-explanatory boot is
3673 * A couple of services gained "systemd-" prefixes in their
3674 name if they wrap systemd code, rather than only external
3675 code. Among them fsck@.service which is now
3676 systemd-fsck@.service.
3678 * The HaveWatchdog property has been removed from the D-Bus
3681 * systemd.confirm_spawn= on the kernel command line should now
3684 * There's a new man page crypttab(5) which details all options
3685 we actually understand.
3687 * systemd-nspawn gained a new --capability= switch to pass
3688 additional capabilities to the container.
3690 * timedated will now read known NTP implementation unit names
3691 from /usr/lib/systemd/ntp-units.d/*.list,
3692 systemd-timedated-ntp.target has been removed.
3694 * journalctl gained a new switch "-b" that lists log data of
3695 the current boot only.
3697 * The notify socket is in the abstract namespace again, in
3698 order to support daemons which chroot() at start-up.
3700 * There is a new Storage= configuration option for journald
3701 which allows configuration of where log data should go. This
3702 also provides a way to disable journal logging entirely, so
3703 that data collected is only forwarded to the console, the
3704 kernel log buffer or another syslog implementation.
3706 * Many bugfixes and optimizations
3708 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3709 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3710 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3711 Shawn Landden, Tom Gundersen
3715 * "systemctl help <unit>" now shows the man page if one is
3718 * Several new man pages have been added.
3720 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3721 MaxLevelConsole= can now be specified in
3722 journald.conf. These options allow reducing the amount of
3723 data stored on disk or forwarded by the log level.
3725 * TimerSlackNSec= can now be specified in system.conf for
3726 PID1. This allows system-wide power savings.
3728 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3729 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3734 * logind is now capable of (optionally) handling power and
3735 sleep keys as well as the lid switch.
3737 * journalctl now understands the syntax "journalctl
3738 /usr/bin/avahi-daemon" to get all log output of a specific
3741 * CapabilityBoundingSet= in system.conf now also influences
3742 the capability bound set of usermode helpers of the kernel.
3744 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3745 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3746 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3747 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3751 * Note that we skipped 139 releases here in order to set the
3752 new version to something that is greater than both udev's
3753 and systemd's most recent version number.
3755 * udev: all udev sources are merged into the systemd source tree now.
3756 All future udev development will happen in the systemd tree. It
3757 is still fully supported to use the udev daemon and tools without
3758 systemd running, like in initramfs or other init systems. Building
3759 udev though, will require the *build* of the systemd tree, but
3760 udev can be properly *run* without systemd.
3762 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3763 should be used to create dead device nodes as workarounds for broken
3766 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3767 no longer supported. udev_monitor_new_from_netlink() needs to be
3768 used to subscribe to events.
3770 * udev: when udevd is started by systemd, processes which are left
3771 behind by forking them off of udev rules, are unconditionally cleaned
3772 up and killed now after the event handling has finished. Services or
3773 daemons must be started as systemd services. Services can be
3774 pulled-in by udev to get started, but they can no longer be directly
3775 forked by udev rules.
3777 * udev: the daemon binary is called systemd-udevd now and installed
3778 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3779 to adapt to that, create symlink, or rename the binary after building
3782 * libudev no longer provides these symbols:
3783 udev_monitor_from_socket()
3784 udev_queue_get_failed_list_entry()
3785 udev_get_{dev,sys,run}_path()
3786 The versions number was bumped and symbol versioning introduced.
3788 * systemd-loginctl and systemd-journalctl have been renamed
3789 to loginctl and journalctl to match systemctl.
3791 * The config files: /etc/systemd/systemd-logind.conf and
3792 /etc/systemd/systemd-journald.conf have been renamed to
3793 logind.conf and journald.conf. Package updates should rename
3794 the files to the new names on upgrade.
3796 * For almost all files the license is now LGPL2.1+, changed
3797 from the previous GPL2.0+. Exceptions are some minor stuff
3798 of udev (which will be changed to LGPL2.1 eventually, too),
3799 and the MIT licensed sd-daemon.[ch] library that is suitable
3800 to be used as drop-in files.
3802 * systemd and logind now handle system sleep states, in
3803 particular suspending and hibernating.
3805 * logind now implements a sleep/shutdown/idle inhibiting logic
3806 suitable for a variety of uses. Soonishly Lennart will blog
3807 about this in more detail.
3809 * var-run.mount and var-lock.mount are no longer provided
3810 (which prevously bind mounted these directories to their new
3811 places). Distributions which have not converted these
3812 directories to symlinks should consider stealing these files
3813 from git history and add them downstream.
3815 * We introduced the Documentation= field for units and added
3816 this to all our shipped units. This is useful to make it
3817 easier to explore the boot and the purpose of the various
3820 * All smaller setup units (such as
3821 systemd-vconsole-setup.service) now detect properly if they
3822 are run in a container and are skipped when
3823 appropriate. This guarantees an entirely noise-free boot in
3824 Linux container environments such as systemd-nspawn.
3826 * A framework for implementing offline system updates is now
3827 integrated, for details see:
3828 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3830 * A new service type Type=idle is available now which helps us
3831 avoiding ugly interleaving of getty output and boot status
3834 * There's now a system-wide CapabilityBoundingSet= option to
3835 globally reduce the set of capabilities for the
3836 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3837 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3838 even CAP_NET_ADMIN system-wide for secure systems.
3840 * There are now system-wide DefaultLimitXXX= options to
3841 globally change the defaults of the various resource limits
3842 for all units started by PID 1.
3844 * Harald Hoyer's systemd test suite has been integrated into
3845 systemd which allows easy testing of systemd builds in qemu
3846 and nspawn. (This is really awesome! Ask us for details!)
3848 * The fstab parser is now implemented as generator, not inside
3851 * systemctl will now warn you if .mount units generated from
3852 /etc/fstab are out of date due to changes in fstab that
3853 have not been read by systemd yet.
3855 * systemd is now suitable for usage in initrds. Dracut has
3856 already been updated to make use of this. With this in place
3857 initrds get a slight bit faster but primarily are much
3858 easier to introspect and debug since "systemctl status" in
3859 the host system can be used to introspect initrd services,
3860 and the journal from the initrd is kept around too.
3862 * systemd-delta has been added, a tool to explore differences
3863 between user/admin configuration and vendor defaults.
3865 * PrivateTmp= now affects both /tmp and /var/tmp.
3867 * Boot time status messages are now much prettier and feature
3868 proper english language. Booting up systemd has never been
3871 * Read-ahead pack files now include the inode number of all
3872 files to pre-cache. When the inode changes the pre-caching
3873 is not attempted. This should be nicer to deal with updated
3874 packages which might result in changes of read-ahead
3877 * We now temporaritly lower the kernel's read_ahead_kb variable
3878 when collecting read-ahead data to ensure the kernel's
3879 built-in read-ahead does not add noise to our measurements
3880 of necessary blocks to pre-cache.
3882 * There's now RequiresMountsFor= to add automatic dependencies
3883 for all mounts necessary for a specific file system path.
3885 * MountAuto= and SwapAuto= have been removed from
3886 system.conf. Mounting file systems at boot has to take place
3889 * nspawn now learned a new switch --uuid= to set the machine
3890 ID on the command line.
3892 * nspawn now learned the -b switch to automatically search
3895 * vt102 is now the default TERM for serial TTYs, upgraded from
3898 * systemd-logind now works on VT-less systems.
3900 * The build tree has been reorganized. The individual
3901 components now have directories of their own.
3903 * A new condition type ConditionPathIsReadWrite= is now available.
3905 * nspawn learned the new -C switch to create cgroups for the
3906 container in other hierarchies.
3908 * We now have support for hardware watchdogs, configurable in
3911 * The scheduled shutdown logic now has a public API.
3913 * We now mount /tmp as tmpfs by default, but this can be
3914 masked and /etc/fstab can override it.
3916 * Since udisks does not make use of /media anymore we are not
3917 mounting a tmpfs on it anymore.
3919 * journalctl gained a new --local switch to only interleave
3920 locally generated journal files.
3922 * We can now load the IMA policy at boot automatically.
3924 * The GTK tools have been split off into a systemd-ui.
3926 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3927 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3928 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3929 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3930 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3931 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3932 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3933 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3934 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3939 * This is mostly a bugfix release
3941 * Support optional initialization of the machine ID from the
3942 KVM or container configured UUID.
3944 * Support immediate reboots with "systemctl reboot -ff"
3946 * Show /etc/os-release data in systemd-analyze output
3948 * Many bugfixes for the journal, including endianness fixes and
3949 ensuring that disk space enforcement works
3951 * sd-login.h is C++ comptaible again
3953 * Extend the /etc/os-release format on request of the Debian
3956 * We now refuse non-UTF8 strings used in various configuration
3957 and unit files. This is done to ensure we do not pass invalid
3958 data over D-Bus or expose it elsewhere.
3960 * Register Mimo USB Screens as suitable for automatic seat
3963 * Read SELinux client context from journal clients in a race
3966 * Reorder configuration file lookup order. /etc now always
3967 overrides /run in order to allow the administrator to always
3968 and unconditionally override vendor supplied or
3969 automatically generated data.
3971 * The various user visible bits of the journal now have man
3972 pages. We still lack man pages for the journal API calls
3975 * We now ship all man pages in HTML format again in the
3978 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3979 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3980 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3985 * This is mostly a bugfix release
3987 * systems lacking /etc/os-release are no longer supported.
3989 * Various functionality updates to libsystemd-login.so
3991 * Track class of PAM logins to distuingish greeters from
3994 Contributions from: Kay Sievers, Lennart Poettering, Michael
3999 * This is an important bugfix release for v41.
4001 * Building man pages is now optional which should be useful
4002 for those building systemd from git but unwilling to install
4005 * Watchdog support for supervising services is now usable. In
4006 a future release support for hardware watchdogs
4007 (i.e. /dev/watchdog) will be added building on this.
4009 * Service start rate limiting is now configurable and can be
4010 turned off per service. When a start rate limit is hit a
4011 reboot can automatically be triggered.
4013 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
4015 Contributions from: Benjamin Franzke, Bill Nottingham,
4016 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
4017 Schmidt, Michał Górny, Piotr Drąg
4021 * The systemd binary is installed /usr/lib/systemd/systemd now;
4022 An existing /sbin/init symlink needs to be adapted with the
4025 * The code that loads kernel modules has been ported to invoke
4026 libkmod directly, instead of modprobe. This means we do not
4027 support systems with module-init-tools anymore.
4029 * Watchdog support is now already useful, but still not
4032 * A new kernel command line option systemd.setenv= is
4033 understood to set system wide environment variables
4034 dynamically at boot.
4036 * We now limit the set of capabilities of systemd-journald.
4038 * We now set SIGPIPE to ignore by default, since it only is
4039 useful in shell pipelines, and has little use in general
4040 code. This can be disabled with IgnoreSIPIPE=no in unit
4043 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
4044 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
4049 * This is mostly a bugfix release
4051 * We now expose the reason why a service failed in the
4052 "Result" D-Bus property.
4054 * Rudimentary service watchdog support (will be completed over
4055 the next few releases.)
4057 * When systemd forks off in order execute some service we will
4058 now immediately changes its argv[0] to reflect which process
4059 it will execute. This is useful to minimize the time window
4060 with a generic argv[0], which makes bootcharts more useful
4062 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
4063 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
4064 Mike Kazantsev, Ray Strode
4068 * This is mostly a test release, but incorporates many
4071 * New systemd-cgtop tool to show control groups by their
4074 * Linking against libacl for ACLs is optional again. If
4075 disabled, support tracking device access for active logins
4076 goes becomes unavailable, and so does access to the user
4077 journals by the respective users.
4079 * If a group "adm" exists, journal files are automatically
4080 owned by them, thus allow members of this group full access
4081 to the system journal as well as all user journals.
4083 * The journal now stores the SELinux context of the logging
4084 client for all entries.
4086 * Add C++ inclusion guards to all public headers
4088 * New output mode "cat" in the journal to print only text
4089 messages, without any meta data like date or time.
4091 * Include tiny X server wrapper as a temporary stop-gap to
4092 teach XOrg udev display enumeration. This is used by display
4093 managers such as gdm, and will go away as soon as XOrg
4094 learned native udev hotplugging for display devices.
4096 * Add new systemd-cat tool for executing arbitrary programs
4097 with STDERR/STDOUT connected to the journal. Can also act as
4098 BSD logger replacement, and does so by default.
4100 * Optionally store all locally generated coredumps in the
4101 journal along with meta data.
4103 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
4104 writing short strings to files (for usage for /sys), and for
4105 creating symlinks, character and block device nodes.
4107 * New unit file option ControlGroupPersistent= to make cgroups
4108 persistent, following the mechanisms outlined in
4109 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
4111 * Support multiple local RTCs in a sane way
4113 * No longer monopolize IO when replaying readahead data on
4114 rotating disks, since we might starve non-file-system IO to
4115 death, since fanotify() will not see accesses done by blkid,
4118 * Do not show kernel threads in systemd-cgls anymore, unless
4119 requested with new -k switch.
4121 Contributions from: Dan Horák, Kay Sievers, Lennart
4122 Poettering, Michal Schmidt
4126 * This is mostly a test release, but incorporates many
4129 * The git repository moved to:
4130 git://anongit.freedesktop.org/systemd/systemd
4131 ssh://git.freedesktop.org/git/systemd/systemd
4133 * First release with the journal
4134 http://0pointer.de/blog/projects/the-journal.html
4136 * The journal replaces both systemd-kmsg-syslogd and
4137 systemd-stdout-bridge.
4139 * New sd_pid_get_unit() API call in libsystemd-logind
4141 * Many systemadm clean-ups
4143 * Introduce remote-fs-pre.target which is ordered before all
4144 remote mounts and may be used to start services before all
4147 * Added Mageia support
4149 * Add bash completion for systemd-loginctl
4151 * Actively monitor PID file creation for daemons which exit in
4152 the parent process before having finished writing the PID
4153 file in the daemon process. Daemons which do this need to be
4154 fixed (i.e. PID file creation must have finished before the
4155 parent exits), but we now react a bit more gracefully to them.
4157 * Add colourful boot output, mimicking the well-known output
4158 of existing distributions.
4160 * New option PassCredentials= for socket units, for
4161 compatibility with a recent kernel ABI breakage.
4163 * /etc/rc.local is now hooked in via a generator binary, and
4164 thus will no longer act as synchronization point during
4167 * systemctl list-unit-files now supports --root=.
4169 * systemd-tmpfiles now understands two new commands: z, Z for
4170 relabelling files according to the SELinux database. This is
4171 useful to apply SELinux labels to specific files in /sys,
4174 * Output of SysV services is now forwarded to both the console
4175 and the journal by default, not only just the console.
4177 * New man pages for all APIs from libsystemd-login.
4179 * The build tree got reorganized and a the build system is a
4180 lot more modular allowing embedded setups to specifically
4181 select the components of systemd they are interested in.
4183 * Support for Linux systems lacking the kernel VT subsystem is
4186 * configure's --with-rootdir= got renamed to
4187 --with-rootprefix= to follow the naming used by udev and
4190 * Unless specified otherwise we will now install to /usr instead
4191 of /usr/local by default.
4193 * Processes with '@' in argv[0][0] are now excluded from the
4194 final shut-down killing spree, following the logic explained
4196 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
4198 * All processes remaining in a service cgroup when we enter
4199 the START or START_PRE states are now killed with
4200 SIGKILL. That means it is no longer possible to spawn
4201 background processes from ExecStart= lines (which was never
4202 supported anyway, and bad style).
4204 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
4205 reloading of units together.
4207 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
4208 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
4209 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
4210 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
4211 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek