1 systemd System and Service Manager
5 * journalctl gained the new options -t/--identifier= to match
6 on the syslog identifier (aka "tag"), as well as --utc to
7 show log timestamps in the UTC timezone. journalctl now also
8 accepts -n/--lines=all to disable line capping in a pager.
10 * Services can notify the manager before they start a reload
11 (by sending RELOADING=1) or shutdown (by sending
12 STOPPING=1). This allows the manager to track and show the
13 internal state of daemons and closes a race condition when
14 the process is still running but has closed its D-Bus
17 * Services with Type=oneshot do not have to have any ExecStart
20 * User units are now loaded also from
21 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
22 /run/systemd/user directory that was already previously
23 supported, but is under the control of the user.
25 * Job timeouts (i.e. time-outs on the time a job that is
26 queued stays in the run queue) can now optionally result in
27 immediate reboot or power-off actions (JobTimeoutAction= and
28 JobTimeoutRebootArgument=). This is useful on ".target"
29 units, to limit the maximum time a target remains
30 undispatched in the run queue, and to trigger an emergency
31 operation in such a case. This is now used by default to
32 turn off the system if boot-up (as defined by everything in
33 basic.target) hangs and does not complete for at least
34 15min. Also, if power-off or reboot hang for at least 30min
35 an immediate power-off/reboot operation is triggered. This
36 functionality is particularly useful to increase reliability
37 on embedded devices, but also on laptops which might
38 accidentally get powered on when carried in a backpack and
39 whose boot stays stuck in a hard disk encryption passphrase
42 * systemd-logind can be configured to also handle lid switch
43 events even when the machine is docked or multiple displays
44 are attached (HandleLidSwitchDocked= option).
46 * A helper binary and a service have been added which can be
47 used to resume from hibernation in the initramfs. A
48 generator will parse the resume= option on the kernel
49 command-line to trigger resume.
51 * A user console daemon systemd-consoled has been
52 added. Currently, it is a preview, and will so far open a
53 single terminal on each session of the user marked as
54 Desktop=systemd-console.
56 * Route metrics can be specified for DHCP routes added by
59 * The SELinux context of socket-actived services can be set
60 from the information provided by the networking stack
61 (SELinuxContextFromNet= option).
63 * Userspace firmware loading support has been removed and
64 the minimum supported kernel version is thus bumped to 3.7.
66 * Timeout for udev workers has been increased from 1 to 3
67 minutes, but a warning will be printed after 1 minute to
68 help diagnose kernel modules that take a long time to load.
70 * Udev rules can now remove tags on devices with TAG-="foobar".
72 * systemd's readahead implementation has been removed. In many
73 circumstatances it didn't give expected benefits even for
74 rotational disk drives and was becoming less relevant in the
75 age of SSDs. As none of the developers has been using
76 rotating media anymore, and nobody stepped up to actively
77 maintain this component of systemd it has now been removed.
79 * Swap units can use Discard= to specify discard options.
80 Discard options specified for swaps in /etc/fstab are now
83 * Docker containers are now detected as a separate type of
86 * The Password Agent protocol gained support for queries where
87 the user input is shown, useful e.g. for usernames.
88 systemd-ask-password gained a new --echo option to turn that
91 * The default sysctl.d/ snippets will now set:
93 net.core.default_qdisc = fq_codel
95 This selects Fair Queueing Controlled Delay as the default
96 queueing discipline for network interfaces. fq_codel helps
97 fight the network bufferbloat problem. It is believed to be
98 a good default with no tuning required for most workloads.
99 Downstream distributions may override this choice. On 10Gbit
100 servers that do not do forwarding, "fq" may perform better.
101 Systems without a good clocksource should use "pfifo_fast".
103 * If kdbus is enabled during build a new option BusPolicy= is
104 available for service units, that allows locking all service
105 processes into a stricter bus policy, in order to limit
106 access to various bus services, or even hide most of them
107 from the service's view entirely.
109 * networkctl will now show the .network and .link file
110 networkd has applied to a specific interface.
112 * sd-login gained a new API call sd_session_get_desktop() to
113 query which desktop environment has been selected for a
116 * UNIX utmp support is now compile-time optional to support
119 * systemctl gained two new commands "add-wants" and
120 "add-requires" for pulling in units from specific targets
123 * If the word "rescue" is specified on the kernel command line
124 the system will now boot into rescue mode (aka
125 rescue.target), which was previously available only by
126 specifying "1" or "systemd.unit=rescue.target" on the kernel
127 command line. This new kernel command line option nicely
128 mirrors the already existing "emergency" kernel command line
131 * New kernel command line options mount.usr=, mount.usrflags=,
132 usrfstype= have been added that match root=, rootflags=,
133 rootfstype= but allow mounting a specific file system to
136 * The $NOTIFY_SOCKET is now also passed to control processesof
137 services, not only the main process.
139 * This version reenables support for fsck's -l switch. This
140 means at least version v2.25 of util-linux is required for
141 operation, otherwise dead-locks on device nodes may
142 occur. Again: you need to update util-linux to at least
143 v2.25 when updating systemd to v217.
145 * The "multi-seat-x" tool has been removed from systemd, as
146 its functionality has been integrated into X servers 1.16,
147 and the tool is hence redundant. It is recommended to update
148 display managers invoking this tool to simply invoke X
149 directly from now on, again.
151 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
152 message flag has been added for all of systemd's PolicyKit
153 authenticated method calls has been added. In particular
154 this now allows optional interactive authorization via
155 PolicyKit for many of PID1's priviliged operations such as
156 unit file enabling and disabling.
158 Contributions from: Andreas Henriksson, Andrei Borzenkov,
159 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
160 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
161 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
162 Herrmann, David Sommerseth, David Strauss, Emil Renner
163 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
164 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
165 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
166 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
167 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
168 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
169 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
170 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
171 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
172 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
173 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
174 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
175 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
176 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
177 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
178 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
181 -- Berlin, 2014-10-28
185 * timedated no longer reads NTP implementation unit names from
186 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
187 implementations should add a
189 Conflicts=systemd-timesyncd.service
191 to their unit files to take over and replace systemd's NTP
192 default functionality.
194 * systemd-sysusers gained a new line type "r" for configuring
195 which UID/GID ranges to allocate system users/groups
196 from. Lines of type "u" may now add an additional column
197 that specifies the home directory for the system user to be
198 created. Also, systemd-sysusers may now optionally read user
199 information from STDIN instead of a file. This is useful for
200 invoking it from RPM preinst scriptlets that need to create
201 users before the first RPM file is installed since these
202 files might need to be owned by them. A new
203 %sysusers_create_inline RPM macro has been introduced to do
204 just that. systemd-sysusers now updates the shadow files as
205 well as the user/group databases, which should enhance
206 compatibility with certain tools like grpck.
208 * A number of bus APIs of PID 1 now optionally consult
209 PolicyKit to permit access for otherwise unprivileged
210 clients under certain conditions. Note that this currently
211 doesn't support interactive authentication yet, but this is
212 expected to be added eventually, too.
214 * /etc/machine-info now has new fields for configuring the
215 deployment environment of the machine, as well as the
216 location of the machine. hostnamectl has been updated with
217 new command to update these fields.
219 * systemd-timesyncd has been updated to automatically acquire
220 NTP server information from systemd-networkd, which might
221 have been discovered via DHCP.
223 * systemd-resolved now includes a caching DNS stub resolver
224 and a complete LLMNR name resolution implementation. A new
225 NSS module "nss-resolve" has been added which can be used
226 instead of glibc's own "nss-dns" to resolve hostnames via
227 systemd-resolved. Hostnames, addresses and arbitrary RRs may
228 be resolved via systemd-resolved D-Bus APIs. In contrast to
229 the glibc internal resolver systemd-resolved is aware of
230 multi-homed system, and keeps DNS server and caches separate
231 and per-interface. Queries are sent simultaneously on all
232 interfaces that have DNS servers configured, in order to
233 properly handle VPNs and local LANs which might resolve
234 separate sets of domain names. systemd-resolved may acquire
235 DNS server information from systemd-networkd automatically,
236 which in turn might have discovered them via DHCP. A tool
237 "systemd-resolve-host" has been added that may be used to
238 query the DNS logic in resolved. systemd-resolved implements
239 IDNA and automatically uses IDNA or UTF-8 encoding depending
240 on whether classic DNS or LLMNR is used as transport. In the
241 next releases we intend to add a DNSSEC and mDNS/DNS-SD
242 implementation to systemd-resolved.
244 * A new NSS module nss-mymachines has been added, that
245 automatically resolves the names of all local registered
246 containers to their respective IP addresses.
248 * A new client tool "networkctl" for systemd-networkd has been
249 added. It currently is entirely passive and will query
250 networking configuration from udev, rtnetlink and networkd,
251 and present it to the user in a very friendly
252 way. Eventually, we hope to extend it to become a full
253 control utility for networkd.
255 * .socket units gained a new DeferAcceptSec= setting that
256 controls the kernels' TCP_DEFER_ACCEPT sockopt for
257 TCP. Similar, support for controlling TCP keep-alive
258 settings has been added (KeepAliveTimeSec=,
259 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
260 turning off Nagle's algorithm on TCP has been added
263 * logind learned a new session type "web", for use in projects
264 like Cockpit which register web clients as PAM sessions.
266 * timer units with at least one OnCalendar= setting will now
267 be started only after timer-sync.target has been
268 reached. This way they will not elapse before the system
269 clock has been corrected by a local NTP client or
270 similar. This is particular useful on RTC-less embedded
271 machines, that come up with an invalid system clock.
273 * systemd-nspawn's --network-veth= switch should now result in
274 stable MAC addresses for both the outer and the inner side
277 * systemd-nspawn gained a new --volatile= switch for running
278 container instances with /etc or /var unpopulated.
280 * The kdbus client code has been updated to use the new Linux
281 3.17 memfd subsystem instead of the old kdbus-specific one.
283 * systemd-networkd's DHCP client and server now support
284 FORCERENEW. There are also new configuration options to
285 configure the vendor client identifier and broadcast mode
288 * systemd will no longer inform the kernel about the current
289 timezone, as this is necessarily incorrect and racy as the
290 kernel has no understanding of DST and similar
291 concepts. This hence means FAT timestamps will be always
292 considered UTC, similar to what Android is already
293 doing. Also, when the RTC is configured to the local time
294 (rather than UTC) systemd will never synchronize back to it,
295 as this might confuse Windows at a later boot.
297 * systemd-analyze gained a new command "verify" for offline
298 validation of unit files.
300 * systemd-networkd gained support for a couple of additional
301 settings for bonding networking setups. Also, the metric for
302 statically configured routes may now be configured. For
303 network interfaces where this is appropriate the peer IP
304 address may now be configured.
306 * systemd-networkd's DHCP client will no longer request
307 broadcasting by default, as this tripped up some networks.
308 For hardware where broadcast is required the feature should
309 be switched back on using RequestBroadcast=yes.
311 * systemd-networkd will now set up IPv4LL addresses (when
312 enabled) even if DHCP is configured successfully.
314 * udev will now default to respect network device names given
315 by the kernel when the kernel indicates that these are
316 predictable. This behavior can be tweaked by changing
317 NamePolicy= in the relevant .link file.
319 * A new library systemd-terminal has been added that
320 implements full TTY stream parsing and rendering. This
321 library is supposed to be used later on for implementing a
322 full userspace VT subsystem, replacing the current kernel
325 * A new tool systemd-journal-upload has been added to push
326 journal data to a remote system running
327 systemd-journal-remote.
329 * journald will no longer forward all local data to another
330 running syslog daemon. This change has been made because
331 rsyslog (which appears to be the most commonly used syslog
332 implementation these days) no longer makes use of this, and
333 instead pulls the data out of the journal on its own. Since
334 forwarding the messages to a non-existent syslog server is
335 more expensive than we assumed we have now turned this
336 off. If you run a syslog server that is not a recent rsyslog
337 version, you have to turn this option on again
338 (ForwardToSyslog= in journald.conf).
340 * journald now optionally supports the LZ4 compressor for
341 larger journal fields. This compressor should perform much
342 better than XZ which was the previous default.
344 * machinectl now shows the IP addresses of local containers,
345 if it knows them, plus the interface name of the container.
347 * A new tool "systemd-escape" has been added that makes it
348 easy to escape strings to build unit names and similar.
350 * sd_notify() messages may now include a new ERRNO= field
351 which is parsed and collected by systemd and shown among the
352 "systemctl status" output for a service.
354 * A new component "systemd-firstboot" has been added that
355 queries the most basic systemd information (timezone,
356 hostname, root password) interactively on first
357 boot. Alternatively it may also be used to provision these
358 things offline on OS images installed into directories.
360 * The default sysctl.d/ snippets will now set
362 net.ipv4.conf.default.promote_secondaries=1
364 This has the benefit of no flushing secondary IP addresses
365 when primary addresses are removed.
367 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
368 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
369 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
370 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
371 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
372 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
373 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
374 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
375 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
376 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
377 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
378 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
379 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
380 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
381 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
383 -- Berlin, 2014-08-19
387 * A new tool systemd-sysusers has been added. This tool
388 creates system users and groups in /etc/passwd and
389 /etc/group, based on static declarative system user/group
390 definitions in /usr/lib/sysusers.d/. This is useful to
391 enable factory resets and volatile systems that boot up with
392 an empty /etc directory, and thus need system users and
393 groups created during early boot. systemd now also ships
394 with two default sysusers.d/ files for the most basic
395 users and groups systemd and the core operating system
398 * A new tmpfiles snippet has been added that rebuilds the
399 essential files in /etc on boot, should they be missing.
401 * A directive for ensuring automatic clean-up of
402 /var/cache/man/ has been removed from the default
403 configuration. This line should now be shipped by the man
404 implementation. The necessary change has been made to the
405 man-db implementation. Note that you need to update your man
406 implementation to one that ships this line, otherwise no
407 automatic clean-up of /var/cache/man will take place.
409 * A new condition ConditionNeedsUpdate= has been added that
410 may conditionalize services to only run when /etc or /var
411 are "older" than the vendor operating system resources in
412 /usr. This is useful for reconstructing or updating /etc
413 after an offline update of /usr or a factory reset, on the
414 next reboot. Services that want to run once after such an
415 update or reset should use this condition and order
416 themselves before the new systemd-update-done.service, which
417 will mark the two directories as fully updated. A number of
418 service files have been added making use of this, to rebuild
419 the udev hardware database, the journald message catalog and
420 dynamic loader cache (ldconfig). The systemd-sysusers tool
421 described above also makes use of this now. With this in
422 place it is now possible to start up a minimal operating
423 system with /etc empty cleanly. For more information on the
424 concepts involved see this recent blog story:
426 http://0pointer.de/blog/projects/stateless.html
428 * A new system group "input" has been introduced, and all
429 input device nodes get this group assigned. This is useful
430 for system-level software to get access to input devices. It
431 complements what is already done for "audio" and "video".
433 * systemd-networkd learnt minimal DHCPv4 server support in
434 addition to the existing DHCPv4 client support. It also
435 learnt DHCPv6 client and IPv6 Router Solicitation client
436 support. The DHCPv4 client gained support for static routes
437 passed in from the server. Note that the [DHCPv4] section
438 known in older systemd-networkd versions has been renamed to
439 [DHCP] and is now also used by the DHCPv6 client. Existing
440 .network files using settings of this section should be
441 updated, though compatibility is maintained. Optionally, the
442 client hostname may now be sent to the DHCP server.
444 * networkd gained support for vxlan virtual networks as well
445 as tun/tap and dummy devices.
447 * networkd gained support for automatic allocation of address
448 ranges for interfaces from a system-wide pool of
449 addresses. This is useful for dynamically managing a large
450 number of interfaces with a single network configuration
451 file. In particular this is useful to easily assign
452 appropriate IP addresses to the veth links of a large number
455 * RPM macros for processing sysusers, sysctl and binfmt
456 drop-in snippets at package installation time have been
459 * The /etc/os-release file should now be placed in
460 /usr/lib/os-release. The old location is automatically
461 created as symlink. /usr/lib is the more appropriate
462 location of this file, since it shall actually describe the
463 vendor operating system shipped in /usr, and not the
464 configuration stored in /etc.
466 * .mount units gained a new boolean SloppyOptions= setting
467 that maps to mount(8)'s -s option which enables permissive
468 parsing of unknown mount options.
470 * tmpfiles learnt a new "L+" directive which creates a symlink
471 but (unlike "L") deletes a pre-existing file first, should
472 it already exist and not already be the correct
473 symlink. Similar, "b+", "c+" and "p+" directives have been
474 added as well, which create block and character devices, as
475 well as fifos in the filesystem, possibly removing any
476 pre-existing files of different types.
478 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
479 'argument' field (which so far specified the source to
480 symlink/copy the files from) is now optional. If omitted the
481 same file os copied from /usr/share/factory/ suffixed by the
482 full destination path. This is useful for populating /etc
483 with essential files, by copying them from vendor defaults
484 shipped in /usr/share/factory/etc.
486 * A new command "systemctl preset-all" has been added that
487 applies the service preset settings to all installed unit
488 files. A new switch --preset-mode= has been added that
489 controls whether only enable or only disable operations
492 * A new command "systemctl is-system-running" has been added
493 that allows checking the overall state of the system, for
494 example whether it is fully up and running.
496 * When the system boots up with an empty /etc, the equivalent
497 to "systemctl preset-all" is executed during early boot, to
498 make sure all default services are enabled after a factory
501 * systemd now contains a minimal preset file that enables the
502 most basic services systemd ships by default.
504 * Unit files' [Install] section gained a new DefaultInstance=
505 field for defining the default instance to create if a
506 template unit is enabled with no instance specified.
508 * A new passive target cryptsetup-pre.target has been added
509 that may be used by services that need to make they run and
510 finish before the first LUKS cryptographic device is set up.
512 * The /dev/loop-control and /dev/btrfs-control device nodes
513 are now owned by the "disk" group by default, opening up
514 access to this group.
516 * systemd-coredump will now automatically generate a
517 stack trace of all core dumps taking place on the system,
518 based on elfutils' libdw library. This stack trace is logged
521 * systemd-coredump may now optionally store coredumps directly
522 on disk (in /var/lib/systemd/coredump, possibly compressed),
523 instead of storing them unconditionally in the journal. This
524 mode is the new default. A new configuration file
525 /etc/systemd/coredump.conf has been added to configure this
526 and other parameters of systemd-coredump.
528 * coredumpctl gained a new "info" verb to show details about a
529 specific coredump. A new switch "-1" has also been added
530 that makes sure to only show information about the most
531 recent entry instead of all entries. Also, as the tool is
532 generally useful now the "systemd-" prefix of the binary
533 name has been removed. Distributions that want to maintain
534 compatibility with the old name should add a symlink from
535 the old name to the new name.
537 * journald's SplitMode= now defaults to "uid". This makes sure
538 that unprivileged users can access their own coredumps with
539 coredumpctl without restrictions.
541 * New kernel command line options "systemd.wants=" (for
542 pulling an additional unit during boot), "systemd.mask="
543 (for masking a specific unit for the boot), and
544 "systemd.debug-shell" (for enabling the debug shell on tty9)
545 have been added. This is implemented in the new generator
546 "systemd-debug-generator".
548 * systemd-nspawn will now by default filter a couple of
549 syscalls for containers, among them those required for
550 kernel module loading, direct x86 IO port access, swap
551 management, and kexec. Most importantly though
552 open_by_handle_at() is now prohibited for containers,
553 closing a hole similar to a recently discussed vulnerability
554 in docker regarding access to files on file hierarchies the
555 container should normally not have access to. Note that for
556 nspawn we generally make no security claims anyway (and
557 this is explicitly documented in the man page), so this is
558 just a fix for one of the most obvious problems.
560 * A new man page file-hierarchy(7) has been added that
561 contains a minimized, modernized version of the file system
562 layout systemd expects, similar in style to the FHS
563 specification or hier(5). A new tool systemd-path(1) has
564 been added to query many of these paths for the local
567 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
568 longer done. Since the directory now has a per-user size
569 limit, and is cleaned on logout this appears unnecessary,
570 in particular since this now brings the lifecycle of this
571 directory closer in line with how IPC objects are handled.
573 * systemd.pc now exports a number of additional directories,
574 including $libdir (which is useful to identify the library
575 path for the primary architecture of the system), and a
576 couple of drop-in directories.
578 * udev's predictable network interface names now use the dev_port
579 sysfs attribute, introduced in linux 3.15 instead of dev_id to
580 distinguish between ports of the same PCI function. dev_id should
581 only be used for ports using the same HW address, hence the need
584 * machined has been updated to export the OS version of a
585 container (read from /etc/os-release and
586 /usr/lib/os-release) on the bus. This is now shown in
587 "machinectl status" for a machine.
589 * A new service setting RestartForceExitStatus= has been
590 added. If configured to a set of exit signals or process
591 return values, the service will be restarted when the main
592 daemon process exits with any of them, regardless of the
595 * systemctl's -H switch for connecting to remote systemd
596 machines has been extended so that it may be used to
597 directly connect to a specific container on the
598 host. "systemctl -H root@foobar:waldi" will now connect as
599 user "root" to host "foobar", and then proceed directly to
600 the container named "waldi". Note that currently you have to
601 authenticate as user "root" for this to work, as entering
602 containers is a privileged operation.
604 Contributions from: Andreas Henriksson, Benjamin Steinwender,
605 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
606 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
607 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
608 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
609 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
610 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
611 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
612 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
613 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
614 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
615 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
617 -- Berlin, 2014-07-03
621 * As an experimental feature, udev now tries to lock the
622 disk device node (flock(LOCK_SH|LOCK_NB)) while it
623 executes events for the disk or any of its partitions.
624 Applications like partitioning programs can lock the
625 disk device node (flock(LOCK_EX)) and claim temporary
626 device ownership that way; udev will entirely skip all event
627 handling for this disk and its partitions. If the disk
628 was opened for writing, the close will trigger a partition
629 table rescan in udev's "watch" facility, and if needed
630 synthesize "change" events for the disk and all its partitions.
631 This is now unconditionally enabled, and if it turns out to
632 cause major problems, we might turn it on only for specific
633 devices, or might need to disable it entirely. Device Mapper
634 devices are excluded from this logic.
636 * We temporarily dropped the "-l" switch for fsck invocations,
637 since they collide with the flock() logic above. util-linux
638 upstream has been changed already to avoid this conflict,
639 and we will readd "-l" as soon as util-linux with this
640 change has been released.
642 * The dependency on libattr has been removed. Since a long
643 time, the extended attribute calls have moved to glibc, and
644 libattr is thus unnecessary.
646 * Virtualization detection works without priviliges now. This
647 means the systemd-detect-virt binary no longer requires
648 CAP_SYS_PTRACE file capabilities, and our daemons can run
649 with fewer privileges.
651 * systemd-networkd now runs under its own "systemd-network"
652 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
653 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
654 loses the ability to write to files owned by root this way.
656 * Similar, systemd-resolved now runs under its own
657 "systemd-resolve" user with no capabilities remaining.
659 * Similar, systemd-bus-proxyd now runs under its own
660 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
662 * systemd-networkd gained support for setting up "veth"
663 virtual ethernet devices for container connectivity, as well
664 as GRE and VTI tunnels.
666 * systemd-networkd will no longer automatically attempt to
667 manually load kernel modules necessary for certain tunnel
668 transports. Instead, it is assumed the kernel loads them
669 automatically when required. This only works correctly on
670 very new kernels. On older kernels, please consider adding
671 the kernel modules to /etc/modules-load.d/ as a work-around.
673 * The resolv.conf file systemd-resolved generates has been
674 moved to /run/systemd/resolve/. If you have a symlink from
675 /etc/resolv.conf, it might be necessary to correct it.
677 * Two new service settings, ProtectHome= and ProtectSystem=,
678 have been added. When enabled, they will make the user data
679 (such as /home) inaccessible or read-only and the system
680 (such as /usr) read-only, for specific services. This allows
681 very light-weight per-service sandboxing to avoid
682 modifications of user data or system files from
683 services. These two new switches have been enabled for all
684 of systemd's long-running services, where appropriate.
686 * Socket units gained new SocketUser= and SocketGroup=
687 settings to set the owner user and group of AF_UNIX sockets
688 and FIFOs in the file system.
690 * Socket units gained a new RemoveOnStop= setting. If enabled,
691 all FIFOS and sockets in the file system will be removed
692 when the specific socket unit is stopped.
694 * Socket units gained a new Symlinks= setting. It takes a list
695 of symlinks to create to file system sockets or FIFOs
696 created by the specific Unix sockets. This is useful to
697 manage symlinks to socket nodes with the same life-cycle as
700 * The /dev/log socket and /dev/initctl FIFO have been moved to
701 /run, and have been replaced by symlinks. This allows
702 connecting to these facilities even if PrivateDevices=yes is
703 used for a service (which makes /dev/log itself unavailable,
704 but /run is left). This also has the benefit of ensuring
705 that /dev only contains device nodes, directories and
706 symlinks, and nothing else.
708 * sd-daemon gained two new calls sd_pid_notify() and
709 sd_pid_notifyf(). They are similar to sd_notify() and
710 sd_notifyf(), but allow overriding of the source PID of
711 notification messages if permissions permit this. This is
712 useful to send notify messages on behalf of a different
713 process (for example, the parent process). The
714 systemd-notify tool has been updated to make use of this
715 when sending messages (so that notification messages now
716 originate from the shell script invoking systemd-notify and
717 not the systemd-notify process itself. This should minimize
718 a race where systemd fails to associate notification
719 messages to services when the originating process already
722 * A new "on-abnormal" setting for Restart= has been added. If
723 set, it will result in automatic restarts on all "abnormal"
724 reasons for a process to exit, which includes unclean
725 signals, core dumps, timeouts and watchdog timeouts, but
726 does not include clean and unclean exit codes or clean
727 signals. Restart=on-abnormal is an alternative for
728 Restart=on-failure for services that shall be able to
729 terminate and avoid restarts on certain errors, by
730 indicating so with an unclean exit code. Restart=on-failure
731 or Restart=on-abnormal is now the recommended setting for
732 all long-running services.
734 * If the InaccessibleDirectories= service setting points to a
735 mount point (or if there are any submounts contained within
736 it), it is now attempted to completely unmount it, to make
737 the file systems truly unavailable for the respective
740 * The ReadOnlyDirectories= service setting and
741 systemd-nspawn's --read-only parameter are now recursively
742 applied to all submounts, too.
744 * Mount units may now be created transiently via the bus APIs.
746 * The support for SysV and LSB init scripts has been removed
747 from the systemd daemon itself. Instead, it is now
748 implemented as a generator that creates native systemd units
749 from these scripts when needed. This enables us to remove a
750 substantial amount of legacy code from PID 1, following the
751 fact that many distributions only ship a very small number
752 of LSB/SysV init scripts nowadays.
754 * Priviliged Xen (dom0) domains are not considered
755 virtualization anymore by the virtualization detection
756 logic. After all, they generally have unrestricted access to
757 the hardware and usually are used to manage the unprivileged
760 * systemd-tmpfiles gained a new "C" line type, for copying
761 files or entire directories.
763 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
764 lines. So far, they have been non-globbing versions of the
765 latter, and have thus been redundant. In future, it is
766 recommended to only use "z". "m" has hence been removed
767 from the documentation, even though it stays supported.
769 * A tmpfiles snippet to recreate the most basic structure in
770 /var has been added. This is enough to create the /var/run →
771 /run symlink and create a couple of structural
772 directories. This allows systems to boot up with an empty or
773 volatile /var. Of course, while with this change, the core OS
774 now is capable with dealing with a volatile /var, not all
775 user services are ready for it. However, we hope that sooner
776 or later, many service daemons will be changed upstream so
777 that they are able to automatically create their necessary
778 directories in /var at boot, should they be missing. This is
779 the first step to allow state-less systems that only require
780 the vendor image for /usr to boot.
782 * systemd-nspawn has gained a new --tmpfs= switch to mount an
783 empty tmpfs instance to a specific directory. This is
784 particularly useful for making use of the automatic
785 reconstruction of /var (see above), by passing --tmpfs=/var.
787 * Access modes specified in tmpfiles snippets may now be
788 prefixed with "~", which indicates that they shall be masked
789 by whether the existing file or directory is currently
790 writable, readable or executable at all. Also, if specified,
791 the sgid/suid/sticky bits will be masked for all
794 * A new passive target unit "network-pre.target" has been
795 added which is useful for services that shall run before any
796 network is configured, for example firewall scripts.
798 * The "floppy" group that previously owned the /dev/fd*
799 devices is no longer used. The "disk" group is now used
800 instead. Distributions should probably deprecate usage of
803 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
804 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
805 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
806 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
807 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
808 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
811 -- Berlin, 2014-06-11
815 * A new "systemd-timesyncd" daemon has been added for
816 synchronizing the system clock across the network. It
817 implements an SNTP client. In contrast to NTP
818 implementations such as chrony or the NTP reference server,
819 this only implements a client side, and does not bother with
820 the full NTP complexity, focusing only on querying time from
821 one remote server and synchronizing the local clock to
822 it. Unless you intend to serve NTP to networked clients or
823 want to connect to local hardware clocks, this simple NTP
824 client should be more than appropriate for most
825 installations. The daemon runs with minimal privileges, and
826 has been hooked up with networkd to only operate when
827 network connectivity is available. The daemon saves the
828 current clock to disk every time a new NTP sync has been
829 acquired, and uses this to possibly correct the system clock
830 early at bootup, in order to accommodate for systems that
831 lack an RTC such as the Raspberry Pi and embedded devices,
832 and to make sure that time monotonically progresses on these
833 systems, even if it is not always correct. To make use of
834 this daemon, a new system user and group "systemd-timesync"
835 needs to be created on installation of systemd.
837 * The queue "seqnum" interface of libudev has been disabled, as
838 it was generally incompatible with device namespacing as
839 sequence numbers of devices go "missing" if the devices are
840 part of a different namespace.
842 * "systemctl list-timers" and "systemctl list-sockets" gained
843 a --recursive switch for showing units of these types also
844 for all local containers, similar in style to the already
845 supported --recursive switch for "systemctl list-units".
847 * A new RebootArgument= setting has been added for service
848 units, which may be used to specify a kernel reboot argument
849 to use when triggering reboots with StartLimitAction=.
851 * A new FailureAction= setting has been added for service
852 units which may be used to specify an operation to trigger
853 when a service fails. This works similarly to
854 StartLimitAction=, but unlike it, controls what is done
855 immediately rather than only after several attempts to
856 restart the service in question.
858 * hostnamed got updated to also expose the kernel name,
859 release, and version on the bus. This is useful for
860 executing commands like hostnamectl with the -H switch.
861 systemd-analyze makes use of this to properly display
862 details when running non-locally.
864 * The bootchart tool can now show cgroup information in the
867 * The CFS CPU quota cgroup attribute is now exposed for
868 services. The new CPUQuota= switch has been added for this
869 which takes a percentage value. Setting this will have the
870 result that a service may never get more CPU time than the
871 specified percentage, even if the machine is otherwise idle.
873 * systemd-networkd learned IPIP and SIT tunnel support.
875 * LSB init scripts exposing a dependency on $network will now
876 get a dependency on network-online.target rather than simply
877 network.target. This should bring LSB handling closer to
878 what it was on SysV systems.
880 * A new fsck.repair= kernel option has been added to control
881 how fsck shall deal with unclean file systems at boot.
883 * The (.ini) configuration file parser will now silently
884 ignore sections whose name begins with "X-". This may be
885 used to maintain application-specific extension sections in unit
888 * machined gained a new API to query the IP addresses of
889 registered containers. "machinectl status" has been updated
890 to show these addresses in its output.
892 * A new call sd_uid_get_display() has been added to the
893 sd-login APIs for querying the "primary" session of a
894 user. The "primary" session of the user is elected from the
895 user's sessions and generally a graphical session is
896 preferred over a text one.
898 * A minimal systemd-resolved daemon has been added. It
899 currently simply acts as a companion to systemd-networkd and
900 manages resolv.conf based on per-interface DNS
901 configuration, possibly supplied via DHCP. In the long run
902 we hope to extend this into a local DNSSEC enabled DNS and
905 * The systemd-networkd-wait-online tool is now enabled by
906 default. It will delay network-online.target until a network
907 connection has been configured. The tool primarily integrates
908 with networkd, but will also make a best effort to make sense
909 of network configuration performed in some other way.
911 * Two new service options StartupCPUShares= and
912 StartupBlockIOWeight= have been added that work similarly to
913 CPUShares= and BlockIOWeight= however only apply during
914 system startup. This is useful to prioritize certain services
915 differently during bootup than during normal runtime.
917 * hostnamed has been changed to prefer the statically
918 configured hostname in /etc/hostname (unless set to
919 'localhost' or empty) over any dynamic one supplied by
920 dhcp. With this change, the rules for picking the hostname
921 match more closely the rules of other configuration settings
922 where the local administrator's configuration in /etc always
923 overrides any other settings.
925 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
926 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
927 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
928 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
929 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
930 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
931 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
932 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
933 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
934 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
935 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
936 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
937 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
938 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
939 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
940 Lindskog, WaLyong Cho, Will Woods, Zbigniew
943 -- Beijing, 2014-05-28
947 * When restoring the screen brightness at boot, stay away from
948 the darkest setting or from the lowest 5% of the available
949 range, depending on which is the larger value of both. This
950 should effectively protect the user from rebooting into a
951 black screen, should the brightness have been set to minimum
954 * sd-login gained a new sd_machine_get_class() call to
955 determine the class ("vm" or "container") of a machine
956 registered with machined.
958 * sd-login gained new calls
959 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
960 to query the identity of the peer of a local AF_UNIX
961 connection. They operate similarly to their sd_pid_get_xyz()
964 * PID 1 will now maintain a system-wide system state engine
965 with the states "starting", "running", "degraded",
966 "maintenance", "stopping". These states are bound to system
967 startup, normal runtime, runtime with at least one failed
968 service, rescue/emergency mode and system shutdown. This
969 state is shown in the "systemctl status" output when no unit
970 name is passed. It is useful to determine system state, in
971 particularly when doing so for many systems or containers at
974 * A new command "list-machines" has been added to "systemctl"
975 that lists all local OS containers and shows their system
976 state (see above), if systemd runs inside of them.
978 * systemctl gained a new "-r" switch to recursively enumerate
979 units on all local containers, when used with the
980 "list-unit" command (which is the default one that is
981 executed when no parameters are specified).
983 * The GPT automatic partition discovery logic will now honour
984 two GPT partition flags: one may be set on a partition to
985 cause it to be mounted read-only, and the other may be set
986 on a partition to ignore it during automatic discovery.
988 * Two new GPT type UUIDs have been added for automatic root
989 partition discovery, for 32-bit and 64-bit ARM. This is not
990 particularly useful for discovering the root directory on
991 these architectures during bare-metal boots (since UEFI is
992 not common there), but still very useful to allow booting of
993 ARM disk images in nspawn with the -i option.
995 * MAC addresses of interfaces created with nspawn's
996 --network-interface= switch will now be generated from the
997 machine name, and thus be stable between multiple invocations
1000 * logind will now automatically remove all IPC objects owned
1001 by a user if she or he fully logs out. This makes sure that
1002 users who are logged out cannot continue to consume IPC
1003 resources. This covers SysV memory, semaphores and message
1004 queues as well as POSIX shared memory and message
1005 queues. Traditionally, SysV and POSIX IPC had no life-cycle
1006 limits. With this functionality, that is corrected. This may
1007 be turned off by using the RemoveIPC= switch of logind.conf.
1009 * The systemd-machine-id-setup and tmpfiles tools gained a
1010 --root= switch to operate on a specific root directory,
1013 * journald can now forward logged messages to the TTYs of all
1014 logged in users ("wall"). This is the default for all
1015 emergency messages now.
1017 * A new tool systemd-journal-remote has been added to stream
1018 journal log messages across the network.
1020 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
1021 controller trees are mounted into it. Note that the
1022 directories mounted beneath it are not read-only. This is a
1023 security measure and is particularly useful because glibc
1024 actually includes a search logic to pick any tmpfs it can
1025 find to implement shm_open() if /dev/shm is not available
1026 (which it might very well be in namespaced setups).
1028 * machinectl gained a new "poweroff" command to cleanly power
1029 down a local OS container.
1031 * The PrivateDevices= unit file setting will now also drop the
1032 CAP_MKNOD capability from the capability bound set, and
1033 imply DevicePolicy=closed.
1035 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
1036 comprehensively on all long-running systemd services where
1037 this is appropriate.
1039 * systemd-udevd will now run in a disassociated mount
1040 namespace. To mount directories from udev rules, make sure to
1041 pull in mount units via SYSTEMD_WANTS properties.
1043 * The kdbus support gained support for uploading policy into
1044 the kernel. sd-bus gained support for creating "monitoring"
1045 connections that can eavesdrop into all bus communication
1046 for debugging purposes.
1048 * Timestamps may now be specified in seconds since the UNIX
1049 epoch Jan 1st, 1970 by specifying "@" followed by the value
1052 * Native tcpwrap support in systemd has been removed. tcpwrap
1053 is old code, not really maintained anymore and has serious
1054 shortcomings, and better options such as firewalls
1055 exist. For setups that require tcpwrap usage, please
1056 consider invoking your socket-activated service via tcpd,
1057 like on traditional inetd.
1059 * A new system.conf configuration option
1060 DefaultTimerAccuracySec= has been added that controls the
1061 default AccuracySec= setting of .timer units.
1063 * Timer units gained a new WakeSystem= switch. If enabled,
1064 timers configured this way will cause the system to resume
1065 from system suspend (if the system supports that, which most
1068 * Timer units gained a new Persistent= switch. If enabled,
1069 timers configured this way will save to disk when they have
1070 been last triggered. This information is then used on next
1071 reboot to possible execute overdue timer events, that
1072 could not take place because the system was powered off.
1073 This enables simple anacron-like behaviour for timer units.
1075 * systemctl's "list-timers" will now also list the time a
1076 timer unit was last triggered in addition to the next time
1077 it will be triggered.
1079 * systemd-networkd will now assign predictable IPv4LL
1080 addresses to its local interfaces.
1082 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
1083 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
1084 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
1085 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
1086 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
1087 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
1088 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
1089 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
1092 -- Berlin, 2014-03-25
1096 * A new unit file setting RestrictAddressFamilies= has been
1097 added to restrict which socket address families unit
1098 processes gain access to. This takes address family names
1099 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1100 attack surface of services via exotic protocol stacks. This
1101 is built on seccomp system call filters.
1103 * Two new unit file settings RuntimeDirectory= and
1104 RuntimeDirectoryMode= have been added that may be used to
1105 manage a per-daemon runtime directories below /run. This is
1106 an alternative for setting up directory permissions with
1107 tmpfiles snippets, and has the advantage that the runtime
1108 directory's lifetime is bound to the daemon runtime and that
1109 the daemon starts up with an empty directory each time. This
1110 is particularly useful when writing services that drop
1111 privileges using the User= or Group= setting.
1113 * The DeviceAllow= unit setting now supports globbing for
1114 matching against device group names.
1116 * The systemd configuration file system.conf gained new
1117 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1118 DefaultMemoryAccounting= to globally turn on/off accounting
1119 for specific resources (cgroups) for all units. These
1120 settings may still be overridden individually in each unit
1123 * systemd-gpt-auto-generator is now able to discover /srv and
1124 root partitions in addition to /home and swap partitions. It
1125 also supports LUKS-encrypted partitions now. With this in
1126 place, automatic discovery of partitions to mount following
1127 the Discoverable Partitions Specification
1128 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1129 is now a lot more complete. This allows booting without
1130 /etc/fstab and without root= on the kernel command line on
1131 systems prepared appropriately.
1133 * systemd-nspawn gained a new --image= switch which allows
1134 booting up disk images and Linux installations on any block
1135 device that follow the Discoverable Partitions Specification
1136 (see above). This means that installations made with
1137 appropriately updated installers may now be started and
1138 deployed using container managers, completely
1139 unmodified. (We hope that libvirt-lxc will add support for
1140 this feature soon, too.)
1142 * systemd-nspawn gained a new --network-macvlan= setting to
1143 set up a private macvlan interface for the
1144 container. Similarly, systemd-networkd gained a new
1145 Kind=macvlan setting in .netdev files.
1147 * systemd-networkd now supports configuring local addresses
1150 * A new tool systemd-network-wait-online has been added to
1151 synchronously wait for network connectivity using
1154 * The sd-bus.h bus API gained a new sd_bus_track object for
1155 tracking the life-cycle of bus peers. Note that sd-bus.h is
1156 still not a public API though (unless you specify
1157 --enable-kdbus on the configure command line, which however
1158 voids your warranty and you get no API stability guarantee).
1160 * The $XDG_RUNTIME_DIR runtime directories for each user are
1161 now individual tmpfs instances, which has the benefit of
1162 introducing separate pools for each user, with individual
1163 size limits, and thus making sure that unprivileged clients
1164 can no longer negatively impact the system or other users by
1165 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
1166 RuntimeDirectorySize= has been introduced that allows
1167 controlling the default size limit for all users. It
1168 defaults to 10% of the available physical memory. This is no
1169 replacement for quotas on tmpfs though (which the kernel
1170 still does not support), as /dev/shm and /tmp are still
1171 shared resources used by both the system and unprivileged
1174 * logind will now automatically turn off automatic suspending
1175 on laptop lid close when more than one display is
1176 connected. This was previously expected to be implemented
1177 individually in desktop environments (such as GNOME),
1178 however has been added to logind now, in order to fix a
1179 boot-time race where a desktop environment might not have
1180 been started yet and thus not been able to take an inhibitor
1181 lock at the time where logind already suspends the system
1182 due to a closed lid.
1184 * logind will now wait at least 30s after each system
1185 suspend/resume cycle, and 3min after system boot before
1186 suspending the system due to a closed laptop lid. This
1187 should give USB docking stations and similar enough time to
1188 be probed and configured after system resume and boot in
1189 order to then act as suspend blocker.
1191 * systemd-run gained a new --property= setting which allows
1192 initialization of resource control properties (and others)
1193 for the created scope or service unit. Example: "systemd-run
1194 --property=BlockIOWeight=10 updatedb" may be used to run
1195 updatedb at a low block IO scheduling weight.
1197 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1198 now also work in --scope mode.
1200 * When systemd is compiled with kdbus support, basic support
1201 for enforced policies is now in place. (Note that enabling
1202 kdbus still voids your warranty and no API compatibility
1205 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1206 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1207 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1208 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1209 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1210 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1211 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1212 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
1213 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
1214 Zbigniew Jędrzejewski-Szmek
1216 -- Berlin, 2014-03-12
1220 * systemd will now relabel /dev after loading the SMACK policy
1221 according to SMACK rules.
1223 * A new unit file option AppArmorProfile= has been added to
1224 set the AppArmor profile for the processes of a unit.
1226 * A new condition check ConditionArchitecture= has been added
1227 to conditionalize units based on the system architecture, as
1228 reported by uname()'s "machine" field.
1230 * systemd-networkd now supports matching on the system
1231 virtualization, architecture, kernel command line, host name
1234 * logind is now a lot more aggressive when suspending the
1235 machine due to a closed laptop lid. Instead of acting only
1236 on the lid close action, it will continuously watch the lid
1237 status and act on it. This is useful for laptops where the
1238 power button is on the outside of the chassis so that it can
1239 be reached without opening the lid (such as the Lenovo
1240 Yoga). On those machines, logind will now immediately
1241 re-suspend the machine if the power button has been
1242 accidentally pressed while the laptop was suspended and in a
1243 backpack or similar.
1245 * logind will now watch SW_DOCK switches and inhibit reaction
1246 to the lid switch if it is pressed. This means that logind
1247 will not suspend the machine anymore if the lid is closed
1248 and the system is docked, if the laptop supports SW_DOCK
1249 notifications via the input layer. Note that ACPI docking
1250 stations do not generate this currently. Also note that this
1251 logic is usually not fully sufficient and Desktop
1252 Environments should take a lid switch inhibitor lock when an
1253 external display is connected, as systemd will not watch
1256 * nspawn will now make use of the devices cgroup controller by
1257 default, and only permit creation of and access to the usual
1258 API device nodes like /dev/null or /dev/random, as well as
1259 access to (but not creation of) the pty devices.
1261 * We will now ship a default .network file for
1262 systemd-networkd that automatically configures DHCP for
1263 network interfaces created by nspawn's --network-veth or
1264 --network-bridge= switches.
1266 * systemd will now understand the usual M, K, G, T suffixes
1267 according to SI conventions (i.e. to the base 1000) when
1268 referring to throughput and hardware metrics. It will stay
1269 with IEC conventions (i.e. to the base 1024) for software
1270 metrics, according to what is customary according to
1271 Wikipedia. We explicitly document which base applies for
1272 each configuration option.
1274 * The DeviceAllow= setting in unit files now supports a syntax
1275 to whitelist an entire group of devices node majors at once,
1276 based on the /proc/devices listing. For example, with the
1277 string "char-pts", it is now possible to whitelist all
1278 current and future pseudo-TTYs at once.
1280 * sd-event learned a new "post" event source. Event sources of
1281 this type are triggered by the dispatching of any event
1282 source of a type that is not "post". This is useful for
1283 implementing clean-up and check event sources that are
1284 triggered by other work being done in the program.
1286 * systemd-networkd is no longer statically enabled, but uses
1287 the usual [Install] sections so that it can be
1288 enabled/disabled using systemctl. It still is enabled by
1291 * When creating a veth interface pair with systemd-nspawn, the
1292 host side will now be prefixed with "vb-" if
1293 --network-bridge= is used, and with "ve-" if --network-veth
1294 is used. This way, it is easy to distinguish these cases on
1295 the host, for example to apply different configuration to
1296 them with systemd-networkd.
1298 * The compatibility libraries for libsystemd-journal.so,
1299 libsystem-id128.so, libsystemd-login.so and
1300 libsystemd-daemon.so do not make use of IFUNC
1301 anymore. Instead, we now build libsystemd.so multiple times
1302 under these alternative names. This means that the footprint
1303 is drastically increased, but given that these are
1304 transitional compatibility libraries, this should not matter
1305 much. This change has been made necessary to support the ARM
1306 platform for these compatibility libraries, as the ARM
1307 toolchain is not really at the same level as the toolchain
1308 for other architectures like x86 and does not support
1309 IFUNC. Please make sure to use --enable-compat-libs only
1310 during a transitional period!
1312 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
1313 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1314 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
1315 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
1316 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
1317 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
1318 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
1319 Zbigniew Jędrzejewski-Szmek
1321 -- Berlin, 2014-02-24
1325 * A new component "systemd-networkd" has been added that can
1326 be used to configure local network interfaces statically or
1327 via DHCP. It is capable of bringing up bridges, VLANs, and
1328 bonding. Currently, no hook-ups for interactive network
1329 configuration are provided. Use this for your initrd,
1330 container, embedded, or server setup if you need a simple,
1331 yet powerful, network configuration solution. This
1332 configuration subsystem is quite nifty, as it allows wildcard
1333 hotplug matching in interfaces. For example, with a single
1334 configuration snippet, you can configure that all Ethernet
1335 interfaces showing up are automatically added to a bridge,
1336 or similar. It supports link-sensing and more.
1338 * A new tool "systemd-socket-proxyd" has been added which can
1339 act as a bidirectional proxy for TCP sockets. This is
1340 useful for adding socket activation support to services that
1341 do not actually support socket activation, including virtual
1342 machines and the like.
1344 * Add a new tool to save/restore rfkill state on
1347 * Save/restore state of keyboard backlights in addition to
1348 display backlights on shutdown/boot.
1350 * udev learned a new SECLABEL{} construct to label device
1351 nodes with a specific security label when they appear. For
1352 now, only SECLABEL{selinux} is supported, but the syntax is
1353 prepared for additional security frameworks.
1355 * udev gained a new scheme to configure link-level attributes
1356 from files in /etc/systemd/network/*.link. These files can
1357 match against MAC address, device path, driver name and type,
1358 and will apply attributes like the naming policy, link speed,
1359 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
1360 address assignment policy (randomized, ...).
1362 * The configuration of network interface naming rules for
1363 "permanent interface names" has changed: a new NamePolicy=
1364 setting in the [Link] section of .link files determines the
1365 priority of possible naming schemes (onboard, slot, mac,
1366 path). The default value of this setting is determined by
1367 /usr/lib/net/links/99-default.link. Old
1368 80-net-name-slot.rules udev configuration file has been
1369 removed, so local configuration overriding this file should
1370 be adapated to override 99-default.link instead.
1372 * When the User= switch is used in a unit file, also
1373 initialize $SHELL= based on the user database entry.
1375 * systemd no longer depends on libdbus. All communication is
1376 now done with sd-bus, systemd's low-level bus library
1379 * kdbus support has been added to PID 1 itself. When kdbus is
1380 enabled, this causes PID 1 to set up the system bus and
1381 enable support for a new ".busname" unit type that
1382 encapsulates bus name activation on kdbus. It works a little
1383 bit like ".socket" units, except for bus names. A new
1384 generator has been added that converts classic dbus1 service
1385 activation files automatically into native systemd .busname
1388 * sd-bus: add a light-weight vtable implementation that allows
1389 defining objects on the bus with a simple static const
1390 vtable array of its methods, signals and properties.
1392 * systemd will not generate or install static dbus
1393 introspection data anymore to /usr/share/dbus-1/interfaces,
1394 as the precise format of these files is unclear, and
1395 nothing makes use of it.
1397 * A proxy daemon is now provided to proxy clients connecting
1398 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1399 compatibility with classic D-Bus.
1401 * A bus driver implementation has been added that supports the
1402 classic D-Bus bus driver calls on kdbus, also for
1403 compatibility purposes.
1405 * A new API "sd-event.h" has been added that implements a
1406 minimal event loop API built around epoll. It provides a
1407 couple of features that direct epoll usage is lacking:
1408 prioritization of events, scales to large numbers of timer
1409 events, per-event timer slack (accuracy), system-wide
1410 coalescing of timer events, exit handlers, watchdog
1411 supervision support using systemd's sd_notify() API, child
1414 * A new API "sd-rntl.h" has been added that provides an API
1415 around the route netlink interface of the kernel, similar in
1416 style to "sd-bus.h".
1418 * A new API "sd-dhcp-client.h" has been added that provides a
1419 small DHCPv4 client-side implementation. This is used by
1422 * There is a new kernel command line option
1423 "systemd.restore_state=0|1". When set to "0", none of the
1424 systemd tools will restore saved runtime state to hardware
1425 devices. More specifically, the rfkill and backlight states
1428 * The FsckPassNo= compatibility option in mount/service units
1429 has been removed. The fstab generator will now add the
1430 necessary dependencies automatically, and does not require
1431 PID1's support for that anymore.
1433 * journalctl gained a new switch, --list-boots, that lists
1434 recent boots with their times and boot IDs.
1436 * The various tools like systemctl, loginctl, timedatectl,
1437 busctl, systemd-run, ... have gained a new switch "-M" to
1438 connect to a specific, local OS container (as direct
1439 connection, without requiring SSH). This works on any
1440 container that is registered with machined, such as those
1441 created by libvirt-lxc or nspawn.
1443 * systemd-run and systemd-analyze also gained support for "-H"
1444 to connect to remote hosts via SSH. This is particularly
1445 useful for systemd-run because it enables queuing of jobs
1446 onto remote systems.
1448 * machinectl gained a new command "login" to open a getty
1449 login in any local container. This works with any container
1450 that is registered with machined (such as those created by
1451 libvirt-lxc or nspawn), and which runs systemd inside.
1453 * machinectl gained a new "reboot" command that may be used to
1454 trigger a reboot on a specific container that is registered
1455 with machined. This works on any container that runs an init
1456 system of some kind.
1458 * systemctl gained a new "list-timers" command to print a nice
1459 listing of installed timer units with the times they elapse
1462 * Alternative reboot() parameters may now be specified on the
1463 "systemctl reboot" command line and are passed to the
1464 reboot() system call.
1466 * systemctl gained a new --job-mode= switch to configure the
1467 mode to queue a job with. This is a more generic version of
1468 --fail, --irreversible, and --ignore-dependencies, which are
1469 still available but not advertised anymore.
1471 * /etc/systemd/system.conf gained new settings to configure
1472 various default timeouts of units, as well as the default
1473 start limit interval and burst. These may still be overridden
1476 * PID1 will now export on the bus profile data of the security
1477 policy upload process (such as the SELinux policy upload to
1480 * journald: when forwarding logs to the console, include
1481 timestamps (following the setting in
1482 /sys/module/printk/parameters/time).
1484 * OnCalendar= in timer units now understands the special
1485 strings "yearly" and "annually". (Both are equivalent)
1487 * The accuracy of timer units is now configurable with the new
1488 AccuracySec= setting. It defaults to 1min.
1490 * A new dependency type JoinsNamespaceOf= has been added that
1491 allows running two services within the same /tmp and network
1492 namespace, if PrivateNetwork= or PrivateTmp= are used.
1494 * A new command "cat" has been added to systemctl. It outputs
1495 the original unit file of a unit, and concatenates the
1496 contents of additional "drop-in" unit file snippets, so that
1497 the full configuration is shown.
1499 * systemctl now supports globbing on the various "list-xyz"
1500 commands, like "list-units" or "list-sockets", as well as on
1501 those commands which take multiple unit names.
1503 * journalctl's --unit= switch gained support for globbing.
1505 * All systemd daemons now make use of the watchdog logic so
1506 that systemd automatically notices when they hang.
1508 * If the $container_ttys environment variable is set,
1509 getty-generator will automatically spawn a getty for each
1510 listed tty. This is useful for container managers to request
1511 login gettys to be spawned on as many ttys as needed.
1513 * %h, %s, %U specifier support is not available anymore when
1514 used in unit files for PID 1. This is because NSS calls are
1515 not safe from PID 1. They stay available for --user
1516 instances of systemd, and as special case for the root user.
1518 * loginctl gained a new "--no-legend" switch to turn off output
1521 * The "sd-login.h" API gained three new calls:
1522 sd_session_is_remote(), sd_session_get_remote_user(),
1523 sd_session_get_remote_host() to query information about
1526 * The udev hardware database now also carries vendor/product
1527 information of SDIO devices.
1529 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
1530 determine whether watchdog notifications are requested by
1533 * Socket-activated per-connection services now include a
1534 short description of the connection parameters in the
1537 * tmpfiles gained a new "--boot" option. When this is not used,
1538 only lines where the command character is not suffixed with
1539 "!" are executed. When this option is specified, those
1540 options are executed too. This partitions tmpfiles
1541 directives into those that can be safely executed at any
1542 time, and those which should be run only at boot (for
1543 example, a line that creates /run/nologin).
1545 * A new API "sd-resolve.h" has been added which provides a simple
1546 asynchronous wrapper around glibc NSS host name resolution
1547 calls, such as getaddrinfo(). In contrast to glibc's
1548 getaddrinfo_a(), it does not use signals. In contrast to most
1549 other asynchronous name resolution libraries, this one does
1550 not reimplement DNS, but reuses NSS, so that alternate
1551 host name resolution systems continue to work, such as mDNS,
1552 LDAP, etc. This API is based on libasyncns, but it has been
1553 cleaned up for inclusion in systemd.
1555 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
1556 "sd-daemon.h" are no longer found in individual libraries
1557 libsystemd-journal.so, libsystemd-login.so,
1558 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
1559 merged them into a single library, libsystemd.so, which
1560 provides all symbols. The reason for this is cyclic
1561 dependencies, as these libraries tend to use each other's
1562 symbols. So far, we have managed to workaround that by linking
1563 a copy of a good part of our code into each of these
1564 libraries again and again, which, however, makes certain
1565 things hard to do, like sharing static variables. Also, it
1566 substantially increases footprint. With this change, there
1567 is only one library for the basic APIs systemd
1568 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
1569 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
1570 library as well, however are subject to the --enable-kdbus
1571 switch (see below). Note that "sd-dhcp-client.h" is not part
1572 of this library (this is because it only consumes, never
1573 provides, services of/to other APIs). To make the transition
1574 easy from the separate libraries to the unified one, we
1575 provide the --enable-compat-libs compile-time switch which
1576 will generate stub libraries that are compatible with the
1577 old ones but redirect all calls to the new one.
1579 * All of the kdbus logic and the new APIs "sd-bus.h",
1580 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
1581 and "sd-utf8.h" are compile-time optional via the
1582 "--enable-kdbus" switch, and they are not compiled in by
1583 default. To make use of kdbus, you have to explicitly enable
1584 the switch. Note however, that neither the kernel nor the
1585 userspace API for all of this is considered stable yet. We
1586 want to maintain the freedom to still change the APIs for
1587 now. By specifying this build-time switch, you acknowledge
1588 that you are aware of the instability of the current
1591 * Also, note that while kdbus is pretty much complete,
1592 it lacks one thing: proper policy support. This means you
1593 can build a fully working system with all features; however,
1594 it will be highly insecure. Policy support will be added in
1595 one of the next releases, at the same time that we will
1596 declare the APIs stable.
1598 * When the kernel command-line argument "kdbus" is specified,
1599 systemd will automatically load the kdbus.ko kernel module. At
1600 this stage of development, it is only useful for testing kdbus
1601 and should not be used in production. Note: if "--enable-kdbus"
1602 is specified, and the kdbus.ko kernel module is available, and
1603 "kdbus" is added to the kernel command line, the entire system
1604 runs with kdbus instead of dbus-daemon, with the above mentioned
1605 problem of missing the system policy enforcement. Also a future
1606 version of kdbus.ko or a newer systemd will not be compatible with
1607 each other, and will unlikely be able to boot the machine if only
1608 one of them is updated.
1610 * systemctl gained a new "import-environment" command which
1611 uploads the caller's environment (or parts thereof) into the
1612 service manager so that it is inherited by services started
1613 by the manager. This is useful to upload variables like
1614 $DISPLAY into the user service manager.
1616 * A new PrivateDevices= switch has been added to service units
1617 which allows running a service with a namespaced /dev
1618 directory that does not contain any device nodes for
1619 physical devices. More specifically, it only includes devices
1620 such as /dev/null, /dev/urandom, and /dev/zero which are API
1623 * logind has been extended to support behaviour like VT
1624 switching on seats that do not support a VT. This makes
1625 multi-session available on seats that are not the first seat
1626 (seat0), and on systems where kernel support for VTs has
1627 been disabled at compile-time.
1629 * If a process holds a delay lock for system sleep or shutdown
1630 and fails to release it in time, we will now log its
1631 identity. This makes it easier to identify processes that
1632 cause slow suspends or power-offs.
1634 * When parsing /etc/crypttab, support for a new key-slot=
1635 option as supported by Debian is added. It allows indicating
1636 which LUKS slot to use on disk, speeding up key loading.
1638 * The sd_journald_sendv() API call has been checked and
1639 officially declared to be async-signal-safe so that it may
1640 be invoked from signal handlers for logging purposes.
1642 * Boot-time status output is now enabled automatically after a
1643 short timeout if boot does not progress, in order to give
1644 the user an indication what she or he is waiting for.
1646 * The boot-time output has been improved to show how much time
1647 remains until jobs expire.
1649 * The KillMode= switch in service units gained a new possible
1650 value "mixed". If set, and the unit is shut down, then the
1651 initial SIGTERM signal is sent only to the main daemon
1652 process, while the following SIGKILL signal is sent to
1653 all remaining processes of the service.
1655 * When a scope unit is registered, a new property "Controller"
1656 may be set. If set to a valid bus name, systemd will send a
1657 RequestStop() signal to this name when it would like to shut
1658 down the scope. This may be used to hook manager logic into
1659 the shutdown logic of scope units. Also, scope units may now
1660 be put in a special "abandoned" state, in which case the
1661 manager process which created them takes no further
1662 responsibilities for it.
1664 * When reading unit files, systemd will now verify
1665 the access mode of these files, and warn about certain
1666 suspicious combinations. This has been added to make it
1667 easier to track down packaging bugs where unit files are
1668 marked executable or world-writable.
1670 * systemd-nspawn gained a new "--setenv=" switch to set
1671 container-wide environment variables. The similar option in
1672 systemd-activate was renamed from "--environment=" to
1673 "--setenv=" for consistency.
1675 * systemd-nspawn has been updated to create a new kdbus domain
1676 for each container that is invoked, thus allowing each
1677 container to have its own set of system and user buses,
1678 independent of the host.
1680 * systemd-nspawn gained a new --drop-capability= switch to run
1681 the container with less capabilities than the default. Both
1682 --drop-capability= and --capability= now take the special
1683 string "all" for dropping or keeping all capabilities.
1685 * systemd-nspawn gained new switches for executing containers
1686 with specific SELinux labels set.
1688 * systemd-nspawn gained a new --quiet switch to not generate
1689 any additional output but the container's own console
1692 * systemd-nspawn gained a new --share-system switch to run a
1693 container without PID namespacing enabled.
1695 * systemd-nspawn gained a new --register= switch to control
1696 whether the container is registered with systemd-machined or
1697 not. This is useful for containers that do not run full
1698 OS images, but only specific apps.
1700 * systemd-nspawn gained a new --keep-unit which may be used
1701 when invoked as the only program from a service unit, and
1702 results in registration of the unit service itself in
1703 systemd-machined, instead of a newly opened scope unit.
1705 * systemd-nspawn gained a new --network-interface= switch for
1706 moving arbitrary interfaces to the container. The new
1707 --network-veth switch creates a virtual Ethernet connection
1708 between host and container. The new --network-bridge=
1709 switch then allows assigning the host side of this virtual
1710 Ethernet connection to a bridge device.
1712 * systemd-nspawn gained a new --personality= switch for
1713 setting the kernel personality for the container. This is
1714 useful when running a 32-bit container on a 64-bit host. A
1715 similar option Personality= is now also available for service
1718 * logind will now also track a "Desktop" identifier for each
1719 session which encodes the desktop environment of it. This is
1720 useful for desktop environments that want to identify
1721 multiple running sessions of itself easily.
1723 * A new SELinuxContext= setting for service units has been
1724 added that allows setting a specific SELinux execution
1725 context for a service.
1727 * Most systemd client tools will now honour $SYSTEMD_LESS for
1728 settings of the "less" pager. By default, these tools will
1729 override $LESS to allow certain operations to work, such as
1730 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1731 influence this logic.
1733 * systemd's "seccomp" hook-up has been changed to make use of
1734 the libseccomp library instead of using its own
1735 implementation. This has benefits for portability among
1738 * For usage together with SystemCallFilter=, a new
1739 SystemCallErrorNumber= setting has been introduced that
1740 allows configuration of a system error number to be returned
1741 on filtered system calls, instead of immediately killing the
1742 process. Also, SystemCallArchitectures= has been added to
1743 limit access to system calls of a particular architecture
1744 (in order to turn off support for unused secondary
1745 architectures). There is also a global
1746 SystemCallArchitectures= setting in system.conf now to turn
1747 off support for non-native system calls system-wide.
1749 * systemd requires a kernel with a working name_to_handle_at(),
1750 please see the kernel config requirements in the README file.
1752 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1753 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1754 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1755 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1756 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1757 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1758 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1759 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1760 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1761 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1762 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1763 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1764 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1765 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1766 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1767 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1768 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1769 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1770 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1771 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1772 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1773 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1774 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1775 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1777 -- Berlin, 2014-02-20
1781 * logind has gained support for facilitating privileged input
1782 and drm device access for unprivileged clients. This work is
1783 useful to allow Wayland display servers (and similar
1784 programs, such as kmscon) to run under the user's ID and
1785 access input and drm devices which are normally
1786 protected. When this is used (and the kernel is new enough)
1787 logind will "mute" IO on the file descriptors passed to
1788 Wayland as long as it is in the background and "unmute" it
1789 if it returns into the foreground. This allows secure
1790 session switching without allowing background sessions to
1791 eavesdrop on input and display data. This also introduces
1792 session switching support if VT support is turned off in the
1793 kernel, and on seats that are not seat0.
1795 * A new kernel command line option luks.options= is understood
1796 now which allows specifying LUKS options for usage for LUKS
1797 encrypted partitions specified with luks.uuid=.
1799 * tmpfiles.d(5) snippets may now use specifier expansion in
1800 path names. More specifically %m, %b, %H, %v, are now
1801 replaced by the local machine id, boot id, hostname, and
1802 kernel version number.
1804 * A new tmpfiles.d(5) command "m" has been introduced which
1805 may be used to change the owner/group/access mode of a file
1806 or directory if it exists, but do nothing if it does not.
1808 * This release removes high-level support for the
1809 MemorySoftLimit= cgroup setting. The underlying kernel
1810 cgroup attribute memory.soft_limit= is currently badly
1811 designed and likely to be removed from the kernel API in its
1812 current form, hence we should not expose it for now.
1814 * The memory.use_hierarchy cgroup attribute is now enabled for
1815 all cgroups systemd creates in the memory cgroup
1816 hierarchy. This option is likely to be come the built-in
1817 default in the kernel anyway, and the non-hierarchial mode
1818 never made much sense in the intrinsically hierarchial
1821 * A new field _SYSTEMD_SLICE= is logged along with all journal
1822 messages containing the slice a message was generated
1823 from. This is useful to allow easy per-customer filtering of
1824 logs among other things.
1826 * systemd-journald will no longer adjust the group of journal
1827 files it creates to the "systemd-journal" group. Instead we
1828 rely on the journal directory to be owned by the
1829 "systemd-journal" group, and its setgid bit set, so that the
1830 kernel file system layer will automatically enforce that
1831 journal files inherit this group assignment. The reason for
1832 this change is that we cannot allow NSS look-ups from
1833 journald which would be necessary to resolve
1834 "systemd-journal" to a numeric GID, because this might
1835 create deadlocks if NSS involves synchronous queries to
1836 other daemons (such as nscd, or sssd) which in turn are
1837 logging clients of journald and might block on it, which
1838 would then dead lock. A tmpfiles.d(5) snippet included in
1839 systemd will make sure the setgid bit and group are
1840 properly set on the journal directory if it exists on every
1841 boot. However, we recommend adjusting it manually after
1842 upgrades too (or from RPM scriptlets), so that the change is
1843 not delayed until next reboot.
1845 * Backlight and random seed files in /var/lib/ have moved into
1846 the /var/lib/systemd/ directory, in order to centralize all
1847 systemd generated files in one directory.
1849 * Boot time performance measurements (as displayed by
1850 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1851 performance information if that's available to determine how
1852 much time BIOS and boot loader initialization required. With
1853 a sufficiently new BIOS you hence no longer need to boot
1854 with Gummiboot to get access to such information.
1856 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1857 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1858 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1859 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1860 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1861 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1862 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1864 -- Berlin, 2013-10-02
1868 * The Restart= option for services now understands a new
1869 on-watchdog setting, which will restart the service
1870 automatically if the service stops sending out watchdog keep
1871 alive messages (as configured with WatchdogSec=).
1873 * The getty generator (which is responsible for bringing up a
1874 getty on configured serial consoles) will no longer only
1875 start a getty on the primary kernel console but on all
1876 others, too. This makes the order in which console= is
1877 specified on the kernel command line less important.
1879 * libsystemd-logind gained a new sd_session_get_vt() call to
1880 retrieve the VT number of a session.
1882 * If the option "tries=0" is set for an entry of /etc/crypttab
1883 its passphrase is queried indefinitely instead of any
1884 maximum number of tries.
1886 * If a service with a configure PID file terminates its PID
1887 file will now be removed automatically if it still exists
1888 afterwards. This should put an end to stale PID files.
1890 * systemd-run will now also take relative binary path names
1891 for execution and no longer insists on absolute paths.
1893 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1894 paths that are optionally prefixed with "-" to indicate that
1895 it should not be considered a failure if they do not exist.
1897 * journalctl -o (and similar commands) now understands a new
1898 output mode "short-precise", it is similar to "short" but
1899 shows timestamps with usec accuracy.
1901 * The option "discard" (as known from Debian) is now
1902 synonymous to "allow-discards" in /etc/crypttab. In fact,
1903 "discard" is preferred now (since it is easier to remember
1906 * Some licensing clean-ups were made, so that more code is now
1907 LGPL-2.1 licensed than before.
1909 * A minimal tool to save/restore the display backlight
1910 brightness across reboots has been added. It will store the
1911 backlight setting as late as possible at shutdown, and
1912 restore it as early as possible during reboot.
1914 * A logic to automatically discover and enable home and swap
1915 partitions on GPT disks has been added. With this in place
1916 /etc/fstab becomes optional for many setups as systemd can
1917 discover certain partitions located on the root disk
1918 automatically. Home partitions are recognized under their
1919 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1920 partitions are recognized under their GPT type ID
1921 0657fd6da4ab43c484e50933c84b4f4f.
1923 * systemd will no longer pass any environment from the kernel
1924 or initrd to system services. If you want to set an
1925 environment for all services, do so via the kernel command
1926 line systemd.setenv= assignment.
1928 * The systemd-sysctl tool no longer natively reads the file
1929 /etc/sysctl.conf. If desired, the file should be symlinked
1930 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1931 legacy support by a symlink rather than built-in code, it
1932 also makes the otherwise hidden order of application of the
1933 different files visible. (Note that this partly reverts to a
1934 pre-198 application order of sysctl knobs!)
1936 * The "systemctl set-log-level" and "systemctl dump" commands
1937 have been moved to systemd-analyze.
1939 * systemd-run learned the new --remain-after-exit switch,
1940 which causes the scope unit not to be cleaned up
1941 automatically after the process terminated.
1943 * tmpfiles learned a new --exclude-prefix= switch to exclude
1944 certain paths from operation.
1946 * journald will now automatically flush all messages to disk
1947 as soon as a message of the log priorities CRIT, ALERT or
1950 Contributions from: Andrew Cook, Brandon Philips, Christian
1951 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1952 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1953 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1954 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1955 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1956 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1957 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1958 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1959 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1960 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1961 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1962 William Giokas, Zbigniew Jędrzejewski-Szmek
1964 -- Berlin, 2013-09-13
1968 * The documentation has been updated to cover the various new
1969 concepts introduced with 205.
1971 * Unit files now understand the new %v specifier which
1972 resolves to the kernel version string as returned by "uname
1975 * systemctl now supports filtering the unit list output by
1976 load state, active state and sub state, using the new
1979 * "systemctl status" will now show the results of the
1980 condition checks (like ConditionPathExists= and similar) of
1981 the last start attempts of the unit. They are also logged to
1984 * "journalctl -b" may now be used to look for boot output of a
1985 specific boot. Try "journalctl -b -1" for the previous boot,
1986 but the syntax is substantially more powerful.
1988 * "journalctl --show-cursor" has been added which prints the
1989 cursor string the last shown log line. This may then be used
1990 with the new "journalctl --after-cursor=" switch to continue
1991 browsing logs from that point on.
1993 * "journalctl --force" may now be used to force regeneration
1996 * Creation of "dead" device nodes has been moved from udev
1997 into kmod and tmpfiles. Previously, udev would read the kmod
1998 databases to pre-generate dead device nodes based on meta
1999 information contained in kernel modules, so that these would
2000 be auto-loaded on access rather then at boot. As this
2001 does not really have much to do with the exposing actual
2002 kernel devices to userspace this has always been slightly
2003 alien in the udev codebase. Following the new scheme kmod
2004 will now generate a runtime snippet for tmpfiles from the
2005 module meta information and it now is tmpfiles' job to the
2006 create the nodes. This also allows overriding access and
2007 other parameters for the nodes using the usual tmpfiles
2008 facilities. As side effect this allows us to remove the
2009 CAP_SYS_MKNOD capability bit from udevd entirely.
2011 * logind's device ACLs may now be applied to these "dead"
2012 devices nodes too, thus finally allowing managed access to
2013 devices such as /dev/snd/sequencer whithout loading the
2014 backing module right-away.
2016 * A new RPM macro has been added that may be used to apply
2017 tmpfiles configuration during package installation.
2019 * systemd-detect-virt and ConditionVirtualization= now can
2020 detect User-Mode-Linux machines (UML).
2022 * journald will now implicitly log the effective capabilities
2023 set of processes in the message metadata.
2025 * systemd-cryptsetup has gained support for TrueCrypt volumes.
2027 * The initrd interface has been simplified (more specifically,
2028 support for passing performance data via environment
2029 variables and fsck results via files in /run has been
2030 removed). These features were non-essential, and are
2031 nowadays available in a much nicer way by having systemd in
2032 the initrd serialize its state and have the hosts systemd
2033 deserialize it again.
2035 * The udev "keymap" data files and tools to apply keyboard
2036 specific mappings of scan to key codes, and force-release
2037 scan code lists have been entirely replaced by a udev
2038 "keyboard" builtin and a hwdb data file.
2040 * systemd will now honour the kernel's "quiet" command line
2041 argument also during late shutdown, resulting in a
2042 completely silent shutdown when used.
2044 * There's now an option to control the SO_REUSEPORT socket
2045 option in .socket units.
2047 * Instance units will now automatically get a per-template
2048 subslice of system.slice unless something else is explicitly
2049 configured. For example, instances of sshd@.service will now
2050 implicitly be placed in system-sshd.slice rather than
2051 system.slice as before.
2053 * Test coverage support may now be enabled at build time.
2055 Contributions from: Dave Reisner, Frederic Crozat, Harald
2056 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
2057 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
2058 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
2059 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
2060 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
2061 Giokas, Zbigniew Jędrzejewski-Szmek
2063 -- Berlin, 2013-07-23
2067 * Two new unit types have been introduced:
2069 Scope units are very similar to service units, however, are
2070 created out of pre-existing processes -- instead of PID 1
2071 forking off the processes. By using scope units it is
2072 possible for system services and applications to group their
2073 own child processes (worker processes) in a powerful way
2074 which then maybe used to organize them, or kill them
2075 together, or apply resource limits on them.
2077 Slice units may be used to partition system resources in an
2078 hierarchial fashion and then assign other units to them. By
2079 default there are now three slices: system.slice (for all
2080 system services), user.slice (for all user sessions),
2081 machine.slice (for VMs and containers).
2083 Slices and scopes have been introduced primarily in
2084 context of the work to move cgroup handling to a
2085 single-writer scheme, where only PID 1
2086 creates/removes/manages cgroups.
2088 * There's a new concept of "transient" units. In contrast to
2089 normal units these units are created via an API at runtime,
2090 not from configuration from disk. More specifically this
2091 means it is now possible to run arbitrary programs as
2092 independent services, with all execution parameters passed
2093 in via bus APIs rather than read from disk. Transient units
2094 make systemd substantially more dynamic then it ever was,
2095 and useful as a general batch manager.
2097 * logind has been updated to make use of scope and slice units
2098 for managing user sessions. As a user logs in he will get
2099 his own private slice unit, to which all sessions are added
2100 as scope units. We also added support for automatically
2101 adding an instance of user@.service for the user into the
2102 slice. Effectively logind will no longer create cgroup
2103 hierarchies on its own now, it will defer entirely to PID 1
2104 for this by means of scope, service and slice units. Since
2105 user sessions this way become entities managed by PID 1
2106 the output of "systemctl" is now a lot more comprehensive.
2108 * A new mini-daemon "systemd-machined" has been added which
2109 may be used by virtualization managers to register local
2110 VMs/containers. nspawn has been updated accordingly, and
2111 libvirt will be updated shortly. machined will collect a bit
2112 of meta information about the VMs/containers, and assign
2113 them their own scope unit (see above). The collected
2114 meta-data is then made available via the "machinectl" tool,
2115 and exposed in "ps" and similar tools. machined/machinectl
2116 is compile-time optional.
2118 * As discussed earlier, the low-level cgroup configuration
2119 options ControlGroup=, ControlGroupModify=,
2120 ControlGroupPersistent=, ControlGroupAttribute= have been
2121 removed. Please use high-level attribute settings instead as
2122 well as slice units.
2124 * A new bus call SetUnitProperties() has been added to alter
2125 various runtime parameters of a unit. This is primarily
2126 useful to alter cgroup parameters dynamically in a nice way,
2127 but will be extended later on to make more properties
2128 modifiable at runtime. systemctl gained a new set-properties
2129 command that wraps this call.
2131 * A new tool "systemd-run" has been added which can be used to
2132 run arbitrary command lines as transient services or scopes,
2133 while configuring a number of settings via the command
2134 line. This tool is currently very basic, however already
2135 very useful. We plan to extend this tool to even allow
2136 queuing of execution jobs with time triggers from the
2137 command line, similar in fashion to "at".
2139 * nspawn will now inform the user explicitly that kernels with
2140 audit enabled break containers, and suggest the user to turn
2143 * Support for detecting the IMA and AppArmor security
2144 frameworks with ConditionSecurity= has been added.
2146 * journalctl gained a new "-k" switch for showing only kernel
2147 messages, mimicking dmesg output; in addition to "--user"
2148 and "--system" switches for showing only user's own logs
2151 * systemd-delta can now show information about drop-in
2152 snippets extending unit files.
2154 * libsystemd-bus has been substantially updated but is still
2155 not available as public API.
2157 * systemd will now look for the "debug" argument on the kernel
2158 command line and enable debug logging, similar to what
2159 "systemd.log_level=debug" already did before.
2161 * "systemctl set-default", "systemctl get-default" has been
2162 added to configure the default.target symlink, which
2163 controls what to boot into by default.
2165 * "systemctl set-log-level" has been added as a convenient
2166 way to raise and lower systemd logging threshold.
2168 * "systemd-analyze plot" will now show the time the various
2169 generators needed for execution, as well as information
2170 about the unit file loading.
2172 * libsystemd-journal gained a new sd_journal_open_files() call
2173 for opening specific journal files. journactl also gained a
2174 new switch to expose this new functionality. Previously we
2175 only supported opening all files from a directory, or all
2176 files from the system, as opening individual files only is
2177 racy due to journal file rotation.
2179 * systemd gained the new DefaultEnvironment= setting in
2180 /etc/systemd/system.conf to set environment variables for
2183 * If a privileged process logs a journal message with the
2184 OBJECT_PID= field set, then journald will automatically
2185 augment this with additional OBJECT_UID=, OBJECT_GID=,
2186 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2187 system services want to log events about specific client
2188 processes. journactl/systemctl has been updated to make use
2189 of this information if all log messages regarding a specific
2192 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2193 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2194 Reisner, David Coppa, David King, David Strauss, Eelco
2195 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2196 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2197 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2198 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2199 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2200 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2201 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2202 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2203 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2204 Łukasz Stelmach, 장동준
2208 * The Python bindings gained some minimal support for the APIs
2209 exposed by libsystemd-logind.
2211 * ConditionSecurity= gained support for detecting SMACK. Since
2212 this condition already supports SELinux and AppArmor we only
2213 miss IMA for this. Patches welcome!
2215 Contributions from: Karol Lewandowski, Lennart Poettering,
2216 Zbigniew Jędrzejewski-Szmek
2220 * systemd-nspawn will now create /etc/resolv.conf if
2221 necessary, before bind-mounting the host's file onto it.
2223 * systemd-nspawn will now store meta information about a
2224 container on the container's cgroup as extended attribute
2225 fields, including the root directory.
2227 * The cgroup hierarchy has been reworked in many ways. All
2228 objects any of the components systemd creates in the cgroup
2229 tree are now suffixed. More specifically, user sessions are
2230 now placed in cgroups suffixed with ".session", users in
2231 cgroups suffixed with ".user", and nspawn containers in
2232 cgroups suffixed with ".nspawn". Furthermore, all cgroup
2233 names are now escaped in a simple scheme to avoid collision
2234 of userspace object names with kernel filenames. This work
2235 is preparation for making these objects relocatable in the
2236 cgroup tree, in order to allow easy resource partitioning of
2237 these objects without causing naming conflicts.
2239 * systemctl list-dependencies gained the new switches
2240 --plain, --reverse, --after and --before.
2242 * systemd-inhibit now shows the process name of processes that
2243 have taken an inhibitor lock.
2245 * nss-myhostname will now also resolve "localhost"
2246 implicitly. This makes /etc/hosts an optional file and
2247 nicely handles that on IPv6 ::1 maps to both "localhost" and
2250 * libsystemd-logind.so gained a new call
2251 sd_get_machine_names() to enumerate running containers and
2252 VMs (currently only supported by very new libvirt and
2253 nspawn). sd_login_monitor can now be used to watch
2254 VMs/containers coming and going.
2256 * .include is not allowed recursively anymore, and only in
2257 unit files. Usually it is better to use drop-in snippets in
2258 .d/*.conf anyway, as introduced with systemd 198.
2260 * systemd-analyze gained a new "critical-chain" command that
2261 determines the slowest chain of units run during system
2262 boot-up. It is very useful for tracking down where
2263 optimizing boot time is the most beneficial.
2265 * systemd will no longer allow manipulating service paths in
2266 the name=systemd:/system cgroup tree using ControlGroup= in
2267 units. (But is still fine with it in all other dirs.)
2269 * There's a new systemd-nspawn@.service service file that may
2270 be used to easily run nspawn containers as system
2271 services. With the container's root directory in
2272 /var/lib/container/foobar it is now sufficient to run
2273 "systemctl start systemd-nspawn@foobar.service" to boot it.
2275 * systemd-cgls gained a new parameter "--machine" to list only
2276 the processes within a certain container.
2278 * ConditionSecurity= now can check for "apparmor". We still
2279 are lacking checks for SMACK and IMA for this condition
2280 check though. Patches welcome!
2282 * A new configuration file /etc/systemd/sleep.conf has been
2283 added that may be used to configure which kernel operation
2284 systemd is supposed to execute when "suspend", "hibernate"
2285 or "hybrid-sleep" is requested. This makes the new kernel
2286 "freeze" state accessible to the user.
2288 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
2289 the passed argument if applicable.
2291 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
2292 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
2293 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
2294 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
2295 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
2296 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
2297 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
2302 * The output of 'systemctl list-jobs' got some polishing. The
2303 '--type=' argument may now be passed more than once. A new
2304 command 'systemctl list-sockets' has been added which shows
2305 a list of kernel sockets systemd is listening on with the
2306 socket units they belong to, plus the units these socket
2309 * The experimental libsystemd-bus library got substantial
2310 updates to work in conjunction with the (also experimental)
2311 kdbus kernel project. It works well enough to exchange
2312 messages with some sophistication. Note that kdbus is not
2313 ready yet, and the library is mostly an elaborate test case
2314 for now, and not installable.
2316 * systemd gained a new unit 'systemd-static-nodes.service'
2317 that generates static device nodes earlier during boot, and
2318 can run in conjunction with udev.
2320 * libsystemd-login gained a new call sd_pid_get_user_unit()
2321 to retrieve the user systemd unit a process is running
2322 in. This is useful for systems where systemd is used as
2325 * systemd-nspawn now places all containers in the new /machine
2326 top-level cgroup directory in the name=systemd
2327 hierarchy. libvirt will soon do the same, so that we get a
2328 uniform separation of /system, /user and /machine for system
2329 services, user processes and containers/virtual
2330 machines. This new cgroup hierarchy is also useful to stick
2331 stable names to specific container instances, which can be
2332 recognized later this way (this name may be controlled
2333 via systemd-nspawn's new -M switch). libsystemd-login also
2334 gained a new call sd_pid_get_machine_name() to retrieve the
2335 name of the container/VM a specific process belongs to.
2337 * bootchart can now store its data in the journal.
2339 * libsystemd-journal gained a new call
2340 sd_journal_add_conjunction() for AND expressions to the
2341 matching logic. This can be used to express more complex
2342 logical expressions.
2344 * journactl can now take multiple --unit= and --user-unit=
2347 * The cryptsetup logic now understands the "luks.key=" kernel
2348 command line switch for specifying a file to read the
2349 decryption key from. Also, if a configured key file is not
2350 found the tool will now automatically fall back to prompting
2353 * Python systemd.journal module was updated to wrap recently
2354 added functions from libsystemd-journal. The interface was
2355 changed to bring the low level interface in s.j._Reader
2356 closer to the C API, and the high level interface in
2357 s.j.Reader was updated to wrap and convert all data about
2360 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
2361 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
2362 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
2363 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
2364 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
2365 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2369 * journalctl --update-catalog now understands a new --root=
2370 option to operate on catalogs found in a different root
2373 * During shutdown after systemd has terminated all running
2374 services a final killing loop kills all remaining left-over
2375 processes. We will now print the name of these processes
2376 when we send SIGKILL to them, since this usually indicates a
2379 * If /etc/crypttab refers to password files stored on
2380 configured mount points automatic dependencies will now be
2381 generated to ensure the specific mount is established first
2382 before the key file is attempted to be read.
2384 * 'systemctl status' will now show information about the
2385 network sockets a socket unit is listening on.
2387 * 'systemctl status' will also shown information about any
2388 drop-in configuration file for units. (Drop-In configuration
2389 files in this context are files such as
2390 /etc/systemd/systemd/foobar.service.d/*.conf)
2392 * systemd-cgtop now optionally shows summed up CPU times of
2393 cgroups. Press '%' while running cgtop to switch between
2394 percentage and absolute mode. This is useful to determine
2395 which cgroups use up the most CPU time over the entire
2396 runtime of the system. systemd-cgtop has also been updated
2397 to be 'pipeable' for processing with further shell tools.
2399 * 'hostnamectl set-hostname' will now allow setting of FQDN
2402 * The formatting and parsing of time span values has been
2403 changed. The parser now understands fractional expressions
2404 such as "5.5h". The formatter will now output fractional
2405 expressions for all time spans under 1min, i.e. "5.123456s"
2406 rather than "5s 123ms 456us". For time spans under 1s
2407 millisecond values are shown, for those under 1ms
2408 microsecond values are shown. This should greatly improve
2409 all time-related output of systemd.
2411 * libsystemd-login and libsystemd-journal gained new
2412 functions for querying the poll() events mask and poll()
2413 timeout value for integration into arbitrary event
2416 * localectl gained the ability to list available X11 keymaps
2417 (models, layouts, variants, options).
2419 * 'systemd-analyze dot' gained the ability to filter for
2420 specific units via shell-style globs, to create smaller,
2421 more useful graphs. I.e. it is now possible to create simple
2422 graphs of all the dependencies between only target units, or
2423 of all units that Avahi has dependencies with.
2425 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2426 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2427 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2428 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2429 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2430 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2431 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2435 * The boot-time readahead implementation for rotating media
2436 will now read the read-ahead data in multiple passes which
2437 consist of all read requests made in equidistant time
2438 intervals. This means instead of strictly reading read-ahead
2439 data in its physical order on disk we now try to find a
2440 middle ground between physical and access time order.
2442 * /etc/os-release files gained a new BUILD_ID= field for usage
2443 on operating systems that provide continuous builds of OS
2446 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
2447 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
2448 William Douglas, Zbigniew Jędrzejewski-Szmek
2452 * systemd-python gained an API exposing libsystemd-daemon.
2454 * The SMACK setup logic gained support for uploading CIPSO
2457 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
2458 ReadOnlyDirectories= and InaccessibleDirectories= has
2459 changed. The private /tmp and /var/tmp directories are now
2460 shared by all processes of a service (which means
2461 ExecStartPre= may now leave data in /tmp that ExecStart= of
2462 the same service can still access). When a service is
2463 stopped its temporary directories are immediately deleted
2464 (normal clean-up with tmpfiles is still done in addition to
2467 * By default, systemd will now set a couple of sysctl
2468 variables in the kernel: the safe sysrq options are turned
2469 on, IP route verification is turned on, and source routing
2470 disabled. The recently added hardlink and softlink
2471 protection of the kernel is turned on. These settings should
2472 be reasonably safe, and good defaults for all new systems.
2474 * The predictable network naming logic may now be turned off
2475 with a new kernel command line switch: net.ifnames=0.
2477 * A new libsystemd-bus module has been added that implements a
2478 pretty complete D-Bus client library. For details see:
2480 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
2482 * journald will now explicitly flush the journal files to disk
2483 at the latest 5min after each write. The file will then also
2484 be marked offline until the next write. This should increase
2485 reliability in case of a crash. The synchronization delay
2486 can be configured via SyncIntervalSec= in journald.conf.
2488 * There's a new remote-fs-setup.target unit that can be used
2489 to pull in specific services when at least one remote file
2490 system is to be mounted.
2492 * There are new targets timers.target and paths.target as
2493 canonical targets to pull user timer and path units in
2494 from. This complements sockets.target with a similar
2495 purpose for socket units.
2497 * libudev gained a new call udev_device_set_attribute_value()
2498 to set sysfs attributes of a device.
2500 * The udev daemon now sets the default number of worker
2501 processes executed in parallel based on the number of available
2502 CPUs instead of the amount of available RAM. This is supposed
2503 to provide a more reliable default and limit a too aggressive
2504 paralellism for setups with 1000s of devices connected.
2506 Contributions from: Auke Kok, Colin Walters, Cristian
2507 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
2508 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
2509 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
2510 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
2511 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
2512 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
2513 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
2514 Zbigniew Jędrzejewski-Szmek
2518 * Configuration of unit files may now be extended via drop-in
2519 files without having to edit/override the unit files
2520 themselves. More specifically, if the administrator wants to
2521 change one value for a service file foobar.service he can
2522 now do so by dropping in a configuration snippet into
2523 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
2524 will load all these snippets and apply them on top of the
2525 main unit configuration file, possibly extending or
2526 overriding its settings. Using these drop-in snippets is
2527 generally nicer than the two earlier options for changing
2528 unit files locally: copying the files from
2529 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
2530 them there; or creating a new file in /etc/systemd/system/
2531 that incorporates the original one via ".include". Drop-in
2532 snippets into these .d/ directories can be placed in any
2533 directory systemd looks for units in, and the usual
2534 overriding semantics between /usr/lib, /etc and /run apply
2537 * Most unit file settings which take lists of items can now be
2538 reset by assigning the empty string to them. For example,
2539 normally, settings such as Environment=FOO=BAR append a new
2540 environment variable assignment to the environment block,
2541 each time they are used. By assigning Environment= the empty
2542 string the environment block can be reset to empty. This is
2543 particularly useful with the .d/*.conf drop-in snippets
2544 mentioned above, since this adds the ability to reset list
2545 settings from vendor unit files via these drop-ins.
2547 * systemctl gained a new "list-dependencies" command for
2548 listing the dependencies of a unit recursively.
2550 * Inhibitors are now honored and listed by "systemctl
2551 suspend", "systemctl poweroff" (and similar) too, not only
2552 GNOME. These commands will also list active sessions by
2555 * Resource limits (as exposed by the various control group
2556 controllers) can now be controlled dynamically at runtime
2557 for all units. More specifically, you can now use a command
2558 like "systemctl set-cgroup-attr foobar.service cpu.shares
2559 2000" to alter the CPU shares a specific service gets. These
2560 settings are stored persistently on disk, and thus allow the
2561 administrator to easily adjust the resource usage of
2562 services with a few simple commands. This dynamic resource
2563 management logic is also available to other programs via the
2564 bus. Almost any kernel cgroup attribute and controller is
2567 * systemd-vconsole-setup will now copy all font settings to
2568 all allocated VTs, where it previously applied them only to
2571 * libsystemd-login gained the new sd_session_get_tty() API
2574 * This release drops support for a few legacy or
2575 distribution-specific LSB facility names when parsing init
2576 scripts: $x-display-manager, $mail-transfer-agent,
2577 $mail-transport-agent, $mail-transfer-agent, $smtp,
2578 $null. Also, the mail-transfer-agent.target unit backing
2579 this has been removed. Distributions which want to retain
2580 compatibility with this should carry the burden for
2581 supporting this themselves and patch support for these back
2582 in, if they really need to. Also, the facilities $syslog and
2583 $local_fs are now ignored, since systemd does not support
2584 early-boot LSB init scripts anymore, and these facilities
2585 are implied anyway for normal services. syslog.target has
2588 * There are new bus calls on PID1's Manager object for
2589 cancelling jobs, and removing snapshot units. Previously,
2590 both calls were only available on the Job and Snapshot
2593 * systemd-journal-gatewayd gained SSL support.
2595 * The various "environment" files, such as /etc/locale.conf
2596 now support continuation lines with a backslash ("\") as
2597 last character in the line, similarly in style (but different)
2598 to how this is supported in shells.
2600 * For normal user processes the _SYSTEMD_USER_UNIT= field is
2601 now implicitly appended to every log entry logged. systemctl
2602 has been updated to filter by this field when operating on a
2603 user systemd instance.
2605 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
2606 CAP_AUDIT_CONTROL capabilities to the capabilities set for
2607 the container. This makes it easier to boot unmodified
2608 Fedora systems in a container, which however still requires
2609 audit=0 to be passed on the kernel command line. Auditing in
2610 kernel and userspace is unfortunately still too broken in
2611 context of containers, hence we recommend compiling it out
2612 of the kernel or using audit=0. Hopefully this will be fixed
2613 one day for good in the kernel.
2615 * nspawn gained the new --bind= and --bind-ro= parameters to
2616 bind mount specific directories from the host into the
2619 * nspawn will now mount its own devpts file system instance
2620 into the container, in order not to leak pty devices from
2621 the host into the container.
2623 * systemd will now read the firmware boot time performance
2624 information from the EFI variables, if the used boot loader
2625 supports this, and takes it into account for boot performance
2626 analysis via "systemd-analyze". This is currently supported
2627 only in conjunction with Gummiboot, but could be supported
2628 by other boot loaders too. For details see:
2630 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
2632 * A new generator has been added that automatically mounts the
2633 EFI System Partition (ESP) to /boot, if that directory
2634 exists, is empty, and no other file system has been
2635 configured to be mounted there.
2637 * logind will now send out PrepareForSleep(false) out
2638 unconditionally, after coming back from suspend. This may be
2639 used by applications as asynchronous notification for
2640 system resume events.
2642 * "systemctl unlock-sessions" has been added, that allows
2643 unlocking the screens of all user sessions at once, similar
2644 to how "systemctl lock-sessions" already locked all users
2645 sessions. This is backed by a new D-Bus call UnlockSessions().
2647 * "loginctl seat-status" will now show the master device of a
2648 seat. (i.e. the device of a seat that needs to be around for
2649 the seat to be considered available, usually the graphics
2652 * tmpfiles gained a new "X" line type, that allows
2653 configuration of files and directories (with wildcards) that
2654 shall be excluded from automatic cleanup ("aging").
2656 * udev default rules set the device node permissions now only
2657 at "add" events, and do not change them any longer with a
2658 later "change" event.
2660 * The log messages for lid events and power/sleep keypresses
2661 now carry a message ID.
2663 * We now have a substantially larger unit test suite, but this
2664 continues to be work in progress.
2666 * udevadm hwdb gained a new --root= parameter to change the
2667 root directory to operate relative to.
2669 * logind will now issue a background sync() request to the kernel
2670 early at shutdown, so that dirty buffers are flushed to disk early
2671 instead of at the last moment, in order to optimize shutdown
2674 * A new bootctl tool has been added that is an interface for
2675 certain boot loader operations. This is currently a preview
2676 and is likely to be extended into a small mechanism daemon
2677 like timedated, localed, hostnamed, and can be used by
2678 graphical UIs to enumerate available boot options, and
2679 request boot into firmware operations.
2681 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2682 the rest of the package. It also has been updated to work
2683 correctly in initrds.
2685 * Policykit previously has been runtime optional, and is now
2686 also compile time optional via a configure switch.
2688 * systemd-analyze has been reimplemented in C. Also "systemctl
2689 dot" has moved into systemd-analyze.
2691 * "systemctl status" with no further parameters will now print
2692 the status of all active or failed units.
2694 * Operations such as "systemctl start" can now be executed
2695 with a new mode "--irreversible" which may be used to queue
2696 operations that cannot accidentally be reversed by a later
2697 job queuing. This is by default used to make shutdown
2698 requests more robust.
2700 * The Python API of systemd now gained a new module for
2701 reading journal files.
2703 * A new tool kernel-install has been added that can install
2704 kernel images according to the Boot Loader Specification:
2706 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2708 * Boot time console output has been improved to provide
2709 animated boot time output for hanging jobs.
2711 * A new tool systemd-activate has been added which can be used
2712 to test socket activation with, directly from the command
2713 line. This should make it much easier to test and debug
2714 socket activation in daemons.
2716 * journalctl gained a new "--reverse" (or -r) option to show
2717 journal output in reverse order (i.e. newest line first).
2719 * journalctl gained a new "--pager-end" (or -e) option to jump
2720 to immediately jump to the end of the journal in the
2721 pager. This is only supported in conjunction with "less".
2723 * journalctl gained a new "--user-unit=" option, that works
2724 similarly to "--unit=" but filters for user units rather than
2727 * A number of unit files to ease adoption of systemd in
2728 initrds has been added. This moves some minimal logic from
2729 the various initrd implementations into systemd proper.
2731 * The journal files are now owned by a new group
2732 "systemd-journal", which exists specifically to allow access
2733 to the journal, and nothing else. Previously, we used the
2734 "adm" group for that, which however possibly covers more
2735 than just journal/log file access. This new group is now
2736 already used by systemd-journal-gatewayd to ensure this
2737 daemon gets access to the journal files and as little else
2738 as possible. Note that "make install" will also set FS ACLs
2739 up for /var/log/journal to give "adm" and "wheel" read
2740 access to it, in addition to "systemd-journal" which owns
2741 the journal files. We recommend that packaging scripts also
2742 add read access to "adm" + "wheel" to /var/log/journal, and
2743 all existing/future journal files. To normal users and
2744 administrators little changes, however packagers need to
2745 ensure to create the "systemd-journal" system group at
2746 package installation time.
2748 * The systemd-journal-gatewayd now runs as unprivileged user
2749 systemd-journal-gateway:systemd-journal-gateway. Packaging
2750 scripts need to create these system user/group at
2753 * timedated now exposes a new boolean property CanNTP that
2754 indicates whether a local NTP service is available or not.
2756 * systemd-detect-virt will now also detect xen PVs
2758 * The pstore file system is now mounted by default, if it is
2761 * In addition to the SELinux and IMA policies we will now also
2762 load SMACK policies at early boot.
2764 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2765 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2766 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2767 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2768 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2769 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2770 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2771 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2772 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2773 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2774 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2775 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2776 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2777 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2781 * Timer units now support calendar time events in addition to
2782 monotonic time events. That means you can now trigger a unit
2783 based on a calendar time specification such as "Thu,Fri
2784 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2785 or fifth day of any month of the year 2013, given that it is
2786 a thursday or friday. This brings timer event support
2787 considerably closer to cron's capabilities. For details on
2788 the supported calendar time specification language see
2791 * udev now supports a number of different naming policies for
2792 network interfaces for predictable names, and a combination
2793 of these policies is now the default. Please see this wiki
2794 document for details:
2796 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2798 * Auke Kok's bootchart implementation has been added to the
2799 systemd tree. It is an optional component that can graph the
2800 boot in quite some detail. It is one of the best bootchart
2801 implementations around and minimal in its code and
2804 * nss-myhostname has been integrated into the systemd source
2805 tree. nss-myhostname guarantees that the local hostname
2806 always stays resolvable via NSS. It has been a weak
2807 requirement of systemd-hostnamed since a long time, and
2808 since its code is actually trivial we decided to just
2809 include it in systemd's source tree. It can be turned off
2810 with a configure switch.
2812 * The read-ahead logic is now capable of properly detecting
2813 whether a btrfs file system is on SSD or rotating media, in
2814 order to optimize the read-ahead scheme. Previously, it was
2815 only capable of detecting this on traditional file systems
2818 * In udev, additional device properties are now read from the
2819 IAB in addition to the OUI database. Also, Bluetooth company
2820 identities are attached to the devices as well.
2822 * In service files %U may be used as specifier that is
2823 replaced by the configured user name of the service.
2825 * nspawn may now be invoked without a controlling TTY. This
2826 makes it suitable for invocation as its own service. This
2827 may be used to set up a simple containerized server system
2828 using only core OS tools.
2830 * systemd and nspawn can now accept socket file descriptors
2831 when they are started for socket activation. This enables
2832 implementation of socket activated nspawn
2833 containers. i.e. think about autospawning an entire OS image
2834 when the first SSH or HTTP connection is received. We expect
2835 that similar functionality will also be added to libvirt-lxc
2838 * journalctl will now suppress ANSI color codes when
2839 presenting log data.
2841 * systemctl will no longer show control group information for
2842 a unit if a the control group is empty anyway.
2844 * logind can now automatically suspend/hibernate/shutdown the
2847 * /etc/machine-info and hostnamed now also expose the chassis
2848 type of the system. This can be used to determine whether
2849 the local system is a laptop, desktop, handset or
2850 tablet. This information may either be configured by the
2851 user/vendor or is automatically determined from ACPI and DMI
2852 information if possible.
2854 * A number of PolicyKit actions are now bound together with
2855 "imply" rules. This should simplify creating UIs because
2856 many actions will now authenticate similar ones as well.
2858 * Unit files learnt a new condition ConditionACPower= which
2859 may be used to conditionalize a unit depending on whether an
2860 AC power source is connected or not, of whether the system
2861 is running on battery power.
2863 * systemctl gained a new "is-failed" verb that may be used in
2864 shell scripts and suchlike to check whether a specific unit
2865 is in the "failed" state.
2867 * The EnvironmentFile= setting in unit files now supports file
2868 globbing, and can hence be used to easily read a number of
2869 environment files at once.
2871 * systemd will no longer detect and recognize specific
2872 distributions. All distribution-specific #ifdeffery has been
2873 removed, systemd is now fully generic and
2874 distribution-agnostic. Effectively, not too much is lost as
2875 a lot of the code is still accessible via explicit configure
2876 switches. However, support for some distribution specific
2877 legacy configuration file formats has been dropped. We
2878 recommend distributions to simply adopt the configuration
2879 files everybody else uses now and convert the old
2880 configuration from packaging scripts. Most distributions
2881 already did that. If that's not possible or desirable,
2882 distributions are welcome to forward port the specific
2883 pieces of code locally from the git history.
2885 * When logging a message about a unit systemd will now always
2886 log the unit name in the message meta data.
2888 * localectl will now also discover system locale data that is
2889 not stored in locale archives, but directly unpacked.
2891 * logind will no longer unconditionally use framebuffer
2892 devices as seat masters, i.e. as devices that are required
2893 to be existing before a seat is considered preset. Instead,
2894 it will now look for all devices that are tagged as
2895 "seat-master" in udev. By default framebuffer devices will
2896 be marked as such, but depending on local systems other
2897 devices might be marked as well. This may be used to
2898 integrate graphics cards using closed source drivers (such
2899 as NVidia ones) more nicely into logind. Note however, that
2900 we recommend using the open source NVidia drivers instead,
2901 and no udev rules for the closed-source drivers will be
2902 shipped from us upstream.
2904 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2905 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2906 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2907 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2908 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2909 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2910 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2911 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2912 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2913 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2914 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2915 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2920 * udev gained support for loading additional device properties
2921 from an indexed database that is keyed by vendor/product IDs
2922 and similar device identifiers. For the beginning this
2923 "hwdb" is populated with data from the well-known PCI and
2924 USB database, but also includes PNP, ACPI and OID data. In
2925 the longer run this indexed database shall grow into
2926 becoming the one central database for non-essential
2927 userspace device metadata. Previously, data from the PCI/USB
2928 database was only attached to select devices, since the
2929 lookup was a relatively expensive operation due to O(n) time
2930 complexity (with n being the number of entries in the
2931 database). Since this is now O(1), we decided to add in this
2932 data for all devices where this is available, by
2933 default. Note that the indexed database needs to be rebuilt
2934 when new data files are installed. To achieve this you need
2935 to update your packaging scripts to invoke "udevadm hwdb
2936 --update" after installation of hwdb data files. For
2937 RPM-based distributions we introduced the new
2938 %udev_hwdb_update macro for this purpose.
2940 * The Journal gained support for the "Message Catalog", an
2941 indexed database to link up additional information with
2942 journal entries. For further details please check:
2944 http://www.freedesktop.org/wiki/Software/systemd/catalog
2946 The indexed message catalog database also needs to be
2947 rebuilt after installation of message catalog files. Use
2948 "journalctl --update-catalog" for this. For RPM-based
2949 distributions we introduced the %journal_catalog_update
2950 macro for this purpose.
2952 * The Python Journal bindings gained support for the standard
2953 Python logging framework.
2955 * The Journal API gained new functions for checking whether
2956 the underlying file system of a journal file is capable of
2957 properly reporting file change notifications, or whether
2958 applications that want to reflect journal changes "live"
2959 need to recheck journal files continuously in appropriate
2962 * It is now possible to set the "age" field for tmpfiles
2963 entries to 0, indicating that files matching this entry
2964 shall always be removed when the directories are cleaned up.
2966 * coredumpctl gained a new "gdb" verb which invokes gdb
2967 right-away on the selected coredump.
2969 * There's now support for "hybrid sleep" on kernels that
2970 support this, in addition to "suspend" and "hibernate". Use
2971 "systemctl hybrid-sleep" to make use of this.
2973 * logind's HandleSuspendKey= setting (and related settings)
2974 now gained support for a new "lock" setting to simply
2975 request the screen lock on all local sessions, instead of
2976 actually executing a suspend or hibernation.
2978 * systemd will now mount the EFI variables file system by
2981 * Socket units now gained support for configuration of the
2982 SMACK security label.
2984 * timedatectl will now output the time of the last and next
2985 daylight saving change.
2987 * We dropped support for various legacy and distro-specific
2988 concepts, such as insserv, early-boot SysV services
2989 (i.e. those for non-standard runlevels such as 'b' or 'S')
2990 or ArchLinux /etc/rc.conf support. We recommend the
2991 distributions who still need support this to either continue
2992 to maintain the necessary patches downstream, or find a
2993 different solution. (Talk to us if you have questions!)
2995 * Various systemd components will now bypass PolicyKit checks
2996 for root and otherwise handle properly if PolicyKit is not
2997 found to be around. This should fix most issues for
2998 PolicyKit-less systems. Quite frankly this should have been
2999 this way since day one. It is absolutely our intention to
3000 make systemd work fine on PolicyKit-less systems, and we
3001 consider it a bug if something does not work as it should if
3002 PolicyKit is not around.
3004 * For embedded systems it is now possible to build udev and
3005 systemd without blkid and/or kmod support.
3007 * "systemctl switch-root" is now capable of switching root
3008 more than once. I.e. in addition to transitions from the
3009 initrd to the host OS it is now possible to transition to
3010 further OS images from the host. This is useful to implement
3011 offline updating tools.
3013 * Various other additions have been made to the RPM macros
3014 shipped with systemd. Use %udev_rules_update() after
3015 installing new udev rules files. %_udevhwdbdir,
3016 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
3017 %_sysctldir are now available which resolve to the right
3018 directories for packages to place various data files in.
3020 * journalctl gained the new --full switch (in addition to
3021 --all, to disable ellipsation for long messages.
3023 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
3024 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
3025 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
3026 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
3027 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
3028 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
3029 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
3030 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
3031 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
3035 * journalctl gained new --since= and --until= switches to
3036 filter by time. It also now supports nice filtering for
3037 units via --unit=/-u.
3039 * Type=oneshot services may use ExecReload= and do the
3042 * The journal daemon now supports time-based rotation and
3043 vacuuming, in addition to the usual disk-space based
3046 * The journal will now index the available field values for
3047 each field name. This enables clients to show pretty drop
3048 downs of available match values when filtering. The bash
3049 completion of journalctl has been updated
3050 accordingly. journalctl gained a new switch -F to list all
3051 values a certain field takes in the journal database.
3053 * More service events are now written as structured messages
3054 to the journal, and made recognizable via message IDs.
3056 * The timedated, localed and hostnamed mini-services which
3057 previously only provided support for changing time, locale
3058 and hostname settings from graphical DEs such as GNOME now
3059 also have a minimal (but very useful) text-based client
3060 utility each. This is probably the nicest way to changing
3061 these settings from the command line now, especially since
3062 it lists available options and is fully integrated with bash
3065 * There's now a new tool "systemd-coredumpctl" to list and
3066 extract coredumps from the journal.
3068 * We now install a README each in /var/log/ and
3069 /etc/rc.d/init.d explaining where the system logs and init
3070 scripts went. This hopefully should help folks who go to
3071 that dirs and look into the otherwise now empty void and
3072 scratch their heads.
3074 * When user-services are invoked (by systemd --user) the
3075 $MANAGERPID env var is set to the PID of systemd.
3077 * SIGRTMIN+24 when sent to a --user instance will now result
3078 in immediate termination of systemd.
3080 * gatewayd received numerous feature additions such as a
3081 "follow" mode, for live syncing and filtering.
3083 * browse.html now allows filtering and showing detailed
3084 information on specific entries. Keyboard navigation and
3085 mouse screen support has been added.
3087 * gatewayd/journalctl now supports HTML5/JSON
3088 Server-Sent-Events as output.
3090 * The SysV init script compatibility logic will now
3091 heuristically determine whether a script supports the
3092 "reload" verb, and only then make this available as
3095 * "systemctl status --follow" has been removed, use "journalctl
3098 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3099 have been removed since they are hardly useful to be
3102 * And I'd like to take the opportunity to specifically mention
3103 Zbigniew for his great contributions. Zbigniew, you rock!
3105 Contributions from: Andrew Eikum, Christian Hesse, Colin
3106 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3107 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3108 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3109 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3110 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3111 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3115 * If /etc/vconsole.conf is non-existent or empty we will no
3116 longer load any console font or key map at boot by
3117 default. Instead the kernel defaults will be left
3118 intact. This is definitely the right thing to do, as no
3119 configuration should mean no configuration, and hard-coding
3120 font names that are different on all archs is probably a bad
3121 idea. Also, the kernel default key map and font should be
3122 good enough for most cases anyway, and mostly identical to
3123 the userspace fonts/key maps we previously overloaded them
3124 with. If distributions want to continue to default to a
3125 non-kernel font or key map they should ship a default
3126 /etc/vconsole.conf with the appropriate contents.
3128 Contributions from: Colin Walters, Daniel J Walsh, Dave
3129 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3130 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3134 * journalctl gained a new --cursor= switch to show entries
3135 starting from the specified location in the journal.
3137 * We now enforce a size limit on journal entry fields exported
3138 with "-o json" in journalctl. Fields larger than 4K will be
3139 assigned null. This can be turned off with --all.
3141 * An (optional) journal gateway daemon is now available as
3142 "systemd-journal-gatewayd.service". This service provides
3143 access to the journal via HTTP and JSON. This functionality
3144 will be used to implement live log synchronization in both
3145 pull and push modes, but has various other users too, such
3146 as easy log access for debugging of embedded devices. Right
3147 now it is already useful to retrieve the journal via HTTP:
3149 # systemctl start systemd-journal-gatewayd.service
3150 # wget http://localhost:19531/entries
3152 This will download the journal contents in a
3153 /var/log/messages compatible format. The same as JSON:
3155 # curl -H"Accept: application/json" http://localhost:19531/entries
3157 This service is also accessible via a web browser where a
3158 single static HTML5 app is served that uses the JSON logic
3159 to enable the user to do some basic browsing of the
3160 journal. This will be extended later on. Here's an example
3161 screenshot of this app in its current state:
3163 http://0pointer.de/public/journal-gatewayd
3165 Contributions from: Kay Sievers, Lennart Poettering, Robert
3166 Milasan, Tom Gundersen
3170 * The bash completion logic is now available for journalctl
3173 * We do not mount the "cpuset" controller anymore together with
3174 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
3175 started if no parameters are assigned to it. "cpuset" hence
3176 broke code that assumed it it could create "cpu" groups and
3179 * journalctl -f will now subscribe to terminal size changes,
3180 and line break accordingly.
3182 Contributions from: Dave Reisner, Kay Sievers, Lennart
3183 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3187 * nspawn will now create a symlink /etc/localtime in the
3188 container environment, copying the host's timezone
3189 setting. Previously this has been done via a bind mount, but
3190 since symlinks cannot be bind mounted this has now been
3191 changed to create/update the appropriate symlink.
3193 * journalctl -n's line number argument is now optional, and
3194 will default to 10 if omitted.
3196 * journald will now log the maximum size the journal files may
3197 take up on disk. This is particularly useful if the default
3198 built-in logic of determining this parameter from the file
3199 system size is used. Use "systemctl status
3200 systemd-journald.service" to see this information.
3202 * The multi-seat X wrapper tool has been stripped down. As X
3203 is now capable of enumerating graphics devices via udev in a
3204 seat-aware way the wrapper is not strictly necessary
3205 anymore. A stripped down temporary stop-gap is still shipped
3206 until the upstream display managers have been updated to
3207 fully support the new X logic. Expect this wrapper to be
3208 removed entirely in one of the next releases.
3210 * HandleSleepKey= in logind.conf has been split up into
3211 HandleSuspendKey= and HandleHibernateKey=. The old setting
3212 is not available anymore. X11 and the kernel are
3213 distuingishing between these keys and we should too. This
3214 also means the inhibition lock for these keys has been split
3217 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
3218 Poettering, Lukas Nykryn, Václav Pavlín
3222 * Whenever a unit changes state we will now log this to the
3223 journal and show along the unit's own log output in
3226 * ConditionPathIsMountPoint= can now properly detect bind
3227 mount points too. (Previously, a bind mount of one file
3228 system to another place in the same file system could not be
3229 detected as mount, since they shared struct stat's st_dev
3232 * We will now mount the cgroup controllers cpu, cpuacct,
3233 cpuset and the controllers net_cls, net_prio together by
3236 * nspawn containers will now have a virtualized boot
3237 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
3238 over with a randomized ID at container initialization). This
3239 has the effect of making "journalctl -b" do the right thing
3242 * The JSON output journal serialization has been updated not
3243 to generate "endless" list objects anymore, but rather one
3244 JSON object per line. This is more in line how most JSON
3245 parsers expect JSON objects. The new output mode
3246 "json-pretty" has been added to provide similar output, but
3247 neatly aligned for readability by humans.
3249 * We dropped all explicit sync() invocations in the shutdown
3250 code. The kernel does this implicitly anyway in the kernel
3251 reboot() syscall. halt(8)'s -n option is now a compatibility
3254 * We now support virtualized reboot() in containers, as
3255 supported by newer kernels. We will fall back to exit() if
3256 CAP_SYS_REBOOT is not available to the container. Also,
3257 nspawn makes use of this now and will actually reboot the
3258 container if the containerized OS asks for that.
3260 * journalctl will only show local log output by default
3261 now. Use --merge (-m) to show remote log output, too.
3263 * libsystemd-journal gained the new sd_journal_get_usage()
3264 call to determine the current disk usage of all journal
3265 files. This is exposed in the new "journalctl --disk-usage"
3268 * journald gained a new configuration setting SplitMode= in
3269 journald.conf which may be used to control how user journals
3270 are split off. See journald.conf(5) for details.
3272 * A new condition type ConditionFileNotEmpty= has been added.
3274 * tmpfiles' "w" lines now support file globbing, to write
3275 multiple files at once.
3277 * We added Python bindings for the journal submission
3278 APIs. More Python APIs for a number of selected APIs will
3279 likely follow. Note that we intend to add native bindings
3280 only for the Python language, as we consider it common
3281 enough to deserve bindings shipped within systemd. There are
3282 various projects outside of systemd that provide bindings
3283 for languages such as PHP or Lua.
3285 * Many conditions will now resolve specifiers such as %i. In
3286 addition, PathChanged= and related directives of .path units
3287 now support specifiers as well.
3289 * There's now a new RPM macro definition for the system preset
3292 * journald will now warn if it ca not forward a message to the
3293 syslog daemon because its socket is full.
3295 * timedated will no longer write or process /etc/timezone,
3296 except on Debian. As we do not support late mounted /usr
3297 anymore /etc/localtime always being a symlink is now safe,
3298 and hence the information in /etc/timezone is not necessary
3301 * logind will now always reserve one VT for a text getty (VT6
3302 by default). Previously if more than 6 X sessions where
3303 started they took up all the VTs with auto-spawned gettys,
3304 so that no text gettys were available anymore.
3306 * udev will now automatically inform the btrfs kernel logic
3307 about btrfs RAID components showing up. This should make
3308 simple hotplug based btrfs RAID assembly work.
3310 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
3311 (but not for its children which will stay at the kernel
3312 default). This should allow setups with a lot more listening
3315 * systemd will now always pass the configured timezone to the
3316 kernel at boot. timedated will do the same when the timezone
3319 * logind's inhibition logic has been updated. By default,
3320 logind will now handle the lid switch, the power and sleep
3321 keys all the time, even in graphical sessions. If DEs want
3322 to handle these events on their own they should take the new
3323 handle-power-key, handle-sleep-key and handle-lid-switch
3324 inhibitors during their runtime. A simple way to achiveve
3325 that is to invoke the DE wrapped in an invocation of:
3327 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
3329 * Access to unit operations is now checked via SELinux taking
3330 the unit file label and client process label into account.
3332 * systemd will now notify the administrator in the journal
3333 when he over-mounts a non-empty directory.
3335 * There are new specifiers that are resolved in unit files,
3336 for the host name (%H), the machine ID (%m) and the boot ID
3339 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
3340 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
3341 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
3342 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
3343 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
3344 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3345 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
3349 * Support for reading structured kernel messages from
3350 /dev/kmsg has now been added and is enabled by default.
3352 * Support for reading kernel messages from /proc/kmsg has now
3353 been removed. If you want kernel messages in the journal
3354 make sure to run a recent kernel (>= 3.5) that supports
3355 reading structured messages from /dev/kmsg (see
3356 above). /proc/kmsg is now exclusive property of classic
3357 syslog daemons again.
3359 * The libudev API gained the new
3360 udev_device_new_from_device_id() call.
3362 * The logic for file system namespace (ReadOnlyDirectory=,
3363 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
3364 require pivot_root() anymore. This means fewer temporary
3365 directories are created below /tmp for this feature.
3367 * nspawn containers will now see and receive all submounts
3368 made on the host OS below the root file system of the
3371 * Forward Secure Sealing is now supported for Journal files,
3372 which provide cryptographical sealing of journal files so
3373 that attackers cannot alter log history anymore without this
3374 being detectable. Lennart will soon post a blog story about
3375 this explaining it in more detail.
3377 * There are two new service settings RestartPreventExitStatus=
3378 and SuccessExitStatus= which allow configuration of exit
3379 status (exit code or signal) which will be excepted from the
3380 restart logic, resp. consider successful.
3382 * journalctl gained the new --verify switch that can be used
3383 to check the integrity of the structure of journal files and
3384 (if Forward Secure Sealing is enabled) the contents of
3387 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3388 and similar symlinks pre-created. This makes running shells
3389 as container init process a lot more fun.
3391 * The fstab support can now handle PARTUUID= and PARTLABEL=
3394 * A new ConditionHost= condition has been added to match
3395 against the hostname (with globs) and machine ID. This is
3396 useful for clusters where a single OS image is used to
3397 provision a large number of hosts which shall run slightly
3398 different sets of services.
3400 * Services which hit the restart limit will now be placed in a
3403 Contributions from: Bertram Poettering, Dave Reisner, Huang
3404 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3405 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3409 * When running in --user mode systemd will now become a
3410 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3411 tree a lot more organized.
3413 * A new PartOf= unit dependency type has been introduced that
3414 may be used to group services in a natural way.
3416 * "systemctl enable" may now be used to enable instances of
3419 * journalctl now prints error log levels in red, and
3420 warning/notice log levels in bright white. It also supports
3421 filtering by log level now.
3423 * cgtop gained a new -n switch (similar to top), to configure
3424 the maximum number of iterations to run for. It also gained
3425 -b, to run in batch mode (accepting no input).
3427 * The suffix ".service" may now be omitted on most systemctl
3428 command lines involving service unit names.
3430 * There's a new bus call in logind to lock all sessions, as
3431 well as a loginctl verb for it "lock-sessions".
3433 * libsystemd-logind.so gained a new call sd_journal_perror()
3434 that works similar to libc perror() but logs to the journal
3435 and encodes structured information about the error number.
3437 * /etc/crypttab entries now understand the new keyfile-size=
3440 * shutdown(8) now can send a (configurable) wall message when
3441 a shutdown is cancelled.
3443 * The mount propagation mode for the root file system will now
3444 default to "shared", which is useful to make containers work
3445 nicely out-of-the-box so that they receive new mounts from
3446 the host. This can be undone locally by running "mount
3447 --make-rprivate /" if needed.
3449 * The prefdm.service file has been removed. Distributions
3450 should maintain this unit downstream if they intend to keep
3451 it around. However, we recommend writing normal unit files
3452 for display managers instead.
3454 * Since systemd is a crucial part of the OS we will now
3455 default to a number of compiler switches that improve
3456 security (hardening) such as read-only relocations, stack
3457 protection, and suchlike.
3459 * The TimeoutSec= setting for services is now split into
3460 TimeoutStartSec= and TimeoutStopSec= to allow configuration
3461 of individual time outs for the start and the stop phase of
3464 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
3465 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
3466 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
3467 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
3468 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
3469 Gundersen, Zbigniew Jędrzejewski-Szmek
3473 * The journal and id128 C APIs are now fully documented as man
3476 * Extra safety checks have been added when transitioning from
3477 the initial RAM disk to the main system to avoid accidental
3480 * /etc/crypttab entries now understand the new keyfile-offset=
3483 * systemctl -t can now be used to filter by unit load state.
3485 * The journal C API gained the new sd_journal_wait() call to
3486 make writing synchronous journal clients easier.
3488 * journalctl gained the new -D switch to show journals from a
3491 * journalctl now displays a special marker between log
3492 messages of two different boots.
3494 * The journal is now explicitly flushed to /var via a service
3495 systemd-journal-flush.service, rather than implicitly simply
3496 by seeing /var/log/journal to be writable.
3498 * journalctl (and the journal C APIs) can now match for much
3499 more complex expressions, with alternatives and
3502 * When transitioning from the initial RAM disk to the main
3503 system we will now kill all processes in a killing spree to
3504 ensure no processes stay around by accident.
3506 * Three new specifiers may be used in unit files: %u, %h, %s
3507 resolve to the user name, user home directory resp. user
3508 shell. This is useful for running systemd user instances.
3510 * We now automatically rotate journal files if their data
3511 object hash table gets a fill level > 75%. We also size the
3512 hash table based on the configured maximum file size. This
3513 together should lower hash collisions drastically and thus
3514 speed things up a bit.
3516 * journalctl gained the new "--header" switch to introspect
3517 header data of journal files.
3519 * A new setting SystemCallFilters= has been added to services
3520 which may be used to apply blacklists or whitelists to
3521 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
3523 * nspawn gained a new --link-journal= switch (and quicker: -j)
3524 to link the container journal with the host. This makes it
3525 very easy to centralize log viewing on the host for all
3526 guests while still keeping the journal files separated.
3528 * Many bugfixes and optimizations
3530 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
3531 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
3532 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
3537 * Several tools now understand kernel command line arguments,
3538 which are only read when run in an initial RAM disk. They
3539 usually follow closely their normal counterparts, but are
3542 * There's a new tool to analyze the readahead files that are
3543 automatically generated at boot. Use:
3545 /usr/lib/systemd/systemd-readahead analyze /.readahead
3547 * We now provide an early debug shell on tty9 if this enabled. Use:
3549 systemctl enable debug-shell.service
3551 * All plymouth related units have been moved into the Plymouth
3552 package. Please make sure to upgrade your Plymouth version
3555 * systemd-tmpfiles now supports getting passed the basename of
3556 a configuration file only, in which case it will look for it
3557 in all appropriate directories automatically.
3559 * udevadm info now takes a /dev or /sys path as argument, and
3560 does the right thing. Example:
3562 udevadm info /dev/sda
3563 udevadm info /sys/class/block/sda
3565 * systemctl now prints a warning if a unit is stopped but a
3566 unit that might trigger it continues to run. Example: a
3567 service is stopped but the socket that activates it is left
3570 * "systemctl status" will now mention if the log output was
3571 shortened due to rotation since a service has been started.
3573 * The journal API now exposes functions to determine the
3574 "cutoff" times due to rotation.
3576 * journald now understands SIGUSR1 and SIGUSR2 for triggering
3577 immediately flushing of runtime logs to /var if possible,
3578 resp. for triggering immediate rotation of the journal
3581 * It is now considered an error if a service is attempted to
3582 be stopped that is not loaded.
3584 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
3586 * systemd-analyze now supports Python 3
3588 * tmpfiles now supports cleaning up directories via aging
3589 where the first level dirs are always kept around but
3590 directories beneath it automatically aged. This is enabled
3591 by prefixing the age field with '~'.
3593 * Seat objects now expose CanGraphical, CanTTY properties
3594 which is required to deal with very fast bootups where the
3595 display manager might be running before the graphics drivers
3596 completed initialization.
3598 * Seat objects now expose a State property.
3600 * We now include RPM macros for service enabling/disabling
3601 based on the preset logic. We recommend RPM based
3602 distributions to make use of these macros if possible. This
3603 makes it simpler to reuse RPM spec files across
3606 * We now make sure that the collected systemd unit name is
3607 always valid when services log to the journal via
3610 * There's a new man page kernel-command-line(7) detailing all
3611 command line options we understand.
3613 * The fstab generator may now be disabled at boot by passing
3614 fstab=0 on the kernel command line.
3616 * A new kernel command line option modules-load= is now understood
3617 to load a specific kernel module statically, early at boot.
3619 * Unit names specified on the systemctl command line are now
3620 automatically escaped as needed. Also, if file system or
3621 device paths are specified they are automatically turned
3622 into the appropriate mount or device unit names. Example:
3624 systemctl status /home
3625 systemctl status /dev/sda
3627 * The SysVConsole= configuration option has been removed from
3628 system.conf parsing.
3630 * The SysV search path is no longer exported on the D-Bus
3633 * The Names= option is been removed from unit file parsing.
3635 * There's a new man page bootup(7) detailing the boot process.
3637 * Every unit and every generator we ship with systemd now
3638 comes with full documentation. The self-explanatory boot is
3641 * A couple of services gained "systemd-" prefixes in their
3642 name if they wrap systemd code, rather than only external
3643 code. Among them fsck@.service which is now
3644 systemd-fsck@.service.
3646 * The HaveWatchdog property has been removed from the D-Bus
3649 * systemd.confirm_spawn= on the kernel command line should now
3652 * There's a new man page crypttab(5) which details all options
3653 we actually understand.
3655 * systemd-nspawn gained a new --capability= switch to pass
3656 additional capabilities to the container.
3658 * timedated will now read known NTP implementation unit names
3659 from /usr/lib/systemd/ntp-units.d/*.list,
3660 systemd-timedated-ntp.target has been removed.
3662 * journalctl gained a new switch "-b" that lists log data of
3663 the current boot only.
3665 * The notify socket is in the abstract namespace again, in
3666 order to support daemons which chroot() at start-up.
3668 * There is a new Storage= configuration option for journald
3669 which allows configuration of where log data should go. This
3670 also provides a way to disable journal logging entirely, so
3671 that data collected is only forwarded to the console, the
3672 kernel log buffer or another syslog implementation.
3674 * Many bugfixes and optimizations
3676 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3677 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3678 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3679 Shawn Landden, Tom Gundersen
3683 * "systemctl help <unit>" now shows the man page if one is
3686 * Several new man pages have been added.
3688 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3689 MaxLevelConsole= can now be specified in
3690 journald.conf. These options allow reducing the amount of
3691 data stored on disk or forwarded by the log level.
3693 * TimerSlackNSec= can now be specified in system.conf for
3694 PID1. This allows system-wide power savings.
3696 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3697 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3702 * logind is now capable of (optionally) handling power and
3703 sleep keys as well as the lid switch.
3705 * journalctl now understands the syntax "journalctl
3706 /usr/bin/avahi-daemon" to get all log output of a specific
3709 * CapabilityBoundingSet= in system.conf now also influences
3710 the capability bound set of usermode helpers of the kernel.
3712 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3713 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3714 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3715 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3719 * Note that we skipped 139 releases here in order to set the
3720 new version to something that is greater than both udev's
3721 and systemd's most recent version number.
3723 * udev: all udev sources are merged into the systemd source tree now.
3724 All future udev development will happen in the systemd tree. It
3725 is still fully supported to use the udev daemon and tools without
3726 systemd running, like in initramfs or other init systems. Building
3727 udev though, will require the *build* of the systemd tree, but
3728 udev can be properly *run* without systemd.
3730 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3731 should be used to create dead device nodes as workarounds for broken
3734 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3735 no longer supported. udev_monitor_new_from_netlink() needs to be
3736 used to subscribe to events.
3738 * udev: when udevd is started by systemd, processes which are left
3739 behind by forking them off of udev rules, are unconditionally cleaned
3740 up and killed now after the event handling has finished. Services or
3741 daemons must be started as systemd services. Services can be
3742 pulled-in by udev to get started, but they can no longer be directly
3743 forked by udev rules.
3745 * udev: the daemon binary is called systemd-udevd now and installed
3746 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3747 to adapt to that, create symlink, or rename the binary after building
3750 * libudev no longer provides these symbols:
3751 udev_monitor_from_socket()
3752 udev_queue_get_failed_list_entry()
3753 udev_get_{dev,sys,run}_path()
3754 The versions number was bumped and symbol versioning introduced.
3756 * systemd-loginctl and systemd-journalctl have been renamed
3757 to loginctl and journalctl to match systemctl.
3759 * The config files: /etc/systemd/systemd-logind.conf and
3760 /etc/systemd/systemd-journald.conf have been renamed to
3761 logind.conf and journald.conf. Package updates should rename
3762 the files to the new names on upgrade.
3764 * For almost all files the license is now LGPL2.1+, changed
3765 from the previous GPL2.0+. Exceptions are some minor stuff
3766 of udev (which will be changed to LGPL2.1 eventually, too),
3767 and the MIT licensed sd-daemon.[ch] library that is suitable
3768 to be used as drop-in files.
3770 * systemd and logind now handle system sleep states, in
3771 particular suspending and hibernating.
3773 * logind now implements a sleep/shutdown/idle inhibiting logic
3774 suitable for a variety of uses. Soonishly Lennart will blog
3775 about this in more detail.
3777 * var-run.mount and var-lock.mount are no longer provided
3778 (which prevously bind mounted these directories to their new
3779 places). Distributions which have not converted these
3780 directories to symlinks should consider stealing these files
3781 from git history and add them downstream.
3783 * We introduced the Documentation= field for units and added
3784 this to all our shipped units. This is useful to make it
3785 easier to explore the boot and the purpose of the various
3788 * All smaller setup units (such as
3789 systemd-vconsole-setup.service) now detect properly if they
3790 are run in a container and are skipped when
3791 appropriate. This guarantees an entirely noise-free boot in
3792 Linux container environments such as systemd-nspawn.
3794 * A framework for implementing offline system updates is now
3795 integrated, for details see:
3796 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3798 * A new service type Type=idle is available now which helps us
3799 avoiding ugly interleaving of getty output and boot status
3802 * There's now a system-wide CapabilityBoundingSet= option to
3803 globally reduce the set of capabilities for the
3804 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3805 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3806 even CAP_NET_ADMIN system-wide for secure systems.
3808 * There are now system-wide DefaultLimitXXX= options to
3809 globally change the defaults of the various resource limits
3810 for all units started by PID 1.
3812 * Harald Hoyer's systemd test suite has been integrated into
3813 systemd which allows easy testing of systemd builds in qemu
3814 and nspawn. (This is really awesome! Ask us for details!)
3816 * The fstab parser is now implemented as generator, not inside
3819 * systemctl will now warn you if .mount units generated from
3820 /etc/fstab are out of date due to changes in fstab that
3821 have not been read by systemd yet.
3823 * systemd is now suitable for usage in initrds. Dracut has
3824 already been updated to make use of this. With this in place
3825 initrds get a slight bit faster but primarily are much
3826 easier to introspect and debug since "systemctl status" in
3827 the host system can be used to introspect initrd services,
3828 and the journal from the initrd is kept around too.
3830 * systemd-delta has been added, a tool to explore differences
3831 between user/admin configuration and vendor defaults.
3833 * PrivateTmp= now affects both /tmp and /var/tmp.
3835 * Boot time status messages are now much prettier and feature
3836 proper english language. Booting up systemd has never been
3839 * Read-ahead pack files now include the inode number of all
3840 files to pre-cache. When the inode changes the pre-caching
3841 is not attempted. This should be nicer to deal with updated
3842 packages which might result in changes of read-ahead
3845 * We now temporaritly lower the kernel's read_ahead_kb variable
3846 when collecting read-ahead data to ensure the kernel's
3847 built-in read-ahead does not add noise to our measurements
3848 of necessary blocks to pre-cache.
3850 * There's now RequiresMountsFor= to add automatic dependencies
3851 for all mounts necessary for a specific file system path.
3853 * MountAuto= and SwapAuto= have been removed from
3854 system.conf. Mounting file systems at boot has to take place
3857 * nspawn now learned a new switch --uuid= to set the machine
3858 ID on the command line.
3860 * nspawn now learned the -b switch to automatically search
3863 * vt102 is now the default TERM for serial TTYs, upgraded from
3866 * systemd-logind now works on VT-less systems.
3868 * The build tree has been reorganized. The individual
3869 components now have directories of their own.
3871 * A new condition type ConditionPathIsReadWrite= is now available.
3873 * nspawn learned the new -C switch to create cgroups for the
3874 container in other hierarchies.
3876 * We now have support for hardware watchdogs, configurable in
3879 * The scheduled shutdown logic now has a public API.
3881 * We now mount /tmp as tmpfs by default, but this can be
3882 masked and /etc/fstab can override it.
3884 * Since udisks does not make use of /media anymore we are not
3885 mounting a tmpfs on it anymore.
3887 * journalctl gained a new --local switch to only interleave
3888 locally generated journal files.
3890 * We can now load the IMA policy at boot automatically.
3892 * The GTK tools have been split off into a systemd-ui.
3894 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3895 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3896 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3897 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3898 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3899 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3900 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3901 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3902 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3907 * This is mostly a bugfix release
3909 * Support optional initialization of the machine ID from the
3910 KVM or container configured UUID.
3912 * Support immediate reboots with "systemctl reboot -ff"
3914 * Show /etc/os-release data in systemd-analyze output
3916 * Many bugfixes for the journal, including endianness fixes and
3917 ensuring that disk space enforcement works
3919 * sd-login.h is C++ comptaible again
3921 * Extend the /etc/os-release format on request of the Debian
3924 * We now refuse non-UTF8 strings used in various configuration
3925 and unit files. This is done to ensure we do not pass invalid
3926 data over D-Bus or expose it elsewhere.
3928 * Register Mimo USB Screens as suitable for automatic seat
3931 * Read SELinux client context from journal clients in a race
3934 * Reorder configuration file lookup order. /etc now always
3935 overrides /run in order to allow the administrator to always
3936 and unconditionally override vendor supplied or
3937 automatically generated data.
3939 * The various user visible bits of the journal now have man
3940 pages. We still lack man pages for the journal API calls
3943 * We now ship all man pages in HTML format again in the
3946 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3947 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3948 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3953 * This is mostly a bugfix release
3955 * systems lacking /etc/os-release are no longer supported.
3957 * Various functionality updates to libsystemd-login.so
3959 * Track class of PAM logins to distuingish greeters from
3962 Contributions from: Kay Sievers, Lennart Poettering, Michael
3967 * This is an important bugfix release for v41.
3969 * Building man pages is now optional which should be useful
3970 for those building systemd from git but unwilling to install
3973 * Watchdog support for supervising services is now usable. In
3974 a future release support for hardware watchdogs
3975 (i.e. /dev/watchdog) will be added building on this.
3977 * Service start rate limiting is now configurable and can be
3978 turned off per service. When a start rate limit is hit a
3979 reboot can automatically be triggered.
3981 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3983 Contributions from: Benjamin Franzke, Bill Nottingham,
3984 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3985 Schmidt, Michał Górny, Piotr Drąg
3989 * The systemd binary is installed /usr/lib/systemd/systemd now;
3990 An existing /sbin/init symlink needs to be adapted with the
3993 * The code that loads kernel modules has been ported to invoke
3994 libkmod directly, instead of modprobe. This means we do not
3995 support systems with module-init-tools anymore.
3997 * Watchdog support is now already useful, but still not
4000 * A new kernel command line option systemd.setenv= is
4001 understood to set system wide environment variables
4002 dynamically at boot.
4004 * We now limit the set of capabilities of systemd-journald.
4006 * We now set SIGPIPE to ignore by default, since it only is
4007 useful in shell pipelines, and has little use in general
4008 code. This can be disabled with IgnoreSIPIPE=no in unit
4011 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
4012 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
4017 * This is mostly a bugfix release
4019 * We now expose the reason why a service failed in the
4020 "Result" D-Bus property.
4022 * Rudimentary service watchdog support (will be completed over
4023 the next few releases.)
4025 * When systemd forks off in order execute some service we will
4026 now immediately changes its argv[0] to reflect which process
4027 it will execute. This is useful to minimize the time window
4028 with a generic argv[0], which makes bootcharts more useful
4030 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
4031 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
4032 Mike Kazantsev, Ray Strode
4036 * This is mostly a test release, but incorporates many
4039 * New systemd-cgtop tool to show control groups by their
4042 * Linking against libacl for ACLs is optional again. If
4043 disabled, support tracking device access for active logins
4044 goes becomes unavailable, and so does access to the user
4045 journals by the respective users.
4047 * If a group "adm" exists, journal files are automatically
4048 owned by them, thus allow members of this group full access
4049 to the system journal as well as all user journals.
4051 * The journal now stores the SELinux context of the logging
4052 client for all entries.
4054 * Add C++ inclusion guards to all public headers
4056 * New output mode "cat" in the journal to print only text
4057 messages, without any meta data like date or time.
4059 * Include tiny X server wrapper as a temporary stop-gap to
4060 teach XOrg udev display enumeration. This is used by display
4061 managers such as gdm, and will go away as soon as XOrg
4062 learned native udev hotplugging for display devices.
4064 * Add new systemd-cat tool for executing arbitrary programs
4065 with STDERR/STDOUT connected to the journal. Can also act as
4066 BSD logger replacement, and does so by default.
4068 * Optionally store all locally generated coredumps in the
4069 journal along with meta data.
4071 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
4072 writing short strings to files (for usage for /sys), and for
4073 creating symlinks, character and block device nodes.
4075 * New unit file option ControlGroupPersistent= to make cgroups
4076 persistent, following the mechanisms outlined in
4077 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
4079 * Support multiple local RTCs in a sane way
4081 * No longer monopolize IO when replaying readahead data on
4082 rotating disks, since we might starve non-file-system IO to
4083 death, since fanotify() will not see accesses done by blkid,
4086 * Do not show kernel threads in systemd-cgls anymore, unless
4087 requested with new -k switch.
4089 Contributions from: Dan Horák, Kay Sievers, Lennart
4090 Poettering, Michal Schmidt
4094 * This is mostly a test release, but incorporates many
4097 * The git repository moved to:
4098 git://anongit.freedesktop.org/systemd/systemd
4099 ssh://git.freedesktop.org/git/systemd/systemd
4101 * First release with the journal
4102 http://0pointer.de/blog/projects/the-journal.html
4104 * The journal replaces both systemd-kmsg-syslogd and
4105 systemd-stdout-bridge.
4107 * New sd_pid_get_unit() API call in libsystemd-logind
4109 * Many systemadm clean-ups
4111 * Introduce remote-fs-pre.target which is ordered before all
4112 remote mounts and may be used to start services before all
4115 * Added Mageia support
4117 * Add bash completion for systemd-loginctl
4119 * Actively monitor PID file creation for daemons which exit in
4120 the parent process before having finished writing the PID
4121 file in the daemon process. Daemons which do this need to be
4122 fixed (i.e. PID file creation must have finished before the
4123 parent exits), but we now react a bit more gracefully to them.
4125 * Add colourful boot output, mimicking the well-known output
4126 of existing distributions.
4128 * New option PassCredentials= for socket units, for
4129 compatibility with a recent kernel ABI breakage.
4131 * /etc/rc.local is now hooked in via a generator binary, and
4132 thus will no longer act as synchronization point during
4135 * systemctl list-unit-files now supports --root=.
4137 * systemd-tmpfiles now understands two new commands: z, Z for
4138 relabelling files according to the SELinux database. This is
4139 useful to apply SELinux labels to specific files in /sys,
4142 * Output of SysV services is now forwarded to both the console
4143 and the journal by default, not only just the console.
4145 * New man pages for all APIs from libsystemd-login.
4147 * The build tree got reorganized and a the build system is a
4148 lot more modular allowing embedded setups to specifically
4149 select the components of systemd they are interested in.
4151 * Support for Linux systems lacking the kernel VT subsystem is
4154 * configure's --with-rootdir= got renamed to
4155 --with-rootprefix= to follow the naming used by udev and
4158 * Unless specified otherwise we will now install to /usr instead
4159 of /usr/local by default.
4161 * Processes with '@' in argv[0][0] are now excluded from the
4162 final shut-down killing spree, following the logic explained
4164 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
4166 * All processes remaining in a service cgroup when we enter
4167 the START or START_PRE states are now killed with
4168 SIGKILL. That means it is no longer possible to spawn
4169 background processes from ExecStart= lines (which was never
4170 supported anyway, and bad style).
4172 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
4173 reloading of units together.
4175 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
4176 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
4177 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
4178 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
4179 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek