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 * systemd-logind can be configured to also handle lid switch
26 events even when the machine is docked or multiple displays
27 are attached (HandleLidSwitchDocked= option).
29 * A helper binary and a service have been added which can be
30 used to resume from hibernation in the initramfs. A
31 generator will parse the resume= option on the kernel
32 command-line to trigger resume.
34 * A user console daemon systemd-consoled has been
35 added. Currently, it is a preview, and will so far open a
36 single terminal on each session of the user marked as
37 Desktop=SYSTEMD-CONSOLE.
39 * Route metrics can be specified for DHCP routes added by
42 * The SELinux context of socket-actived services can be set
43 from the information provided by the networking stack
44 (SELinuxContextFromNet= option).
46 * Userspace firmware loading support has been removed and
47 the minimum supported kernel version is thus bumped to 3.7.
49 * Timeout for udev workers has been increased from 1 to 3
50 minutes, but a warning will be printed after 1 minute to
51 help diagnose kernel modules that take a long time to load.
53 * Udev rules can now remove tags on devices with TAG-="foobar".
55 * systemd's readahead implementation has been removed. In many
56 circumstatances it didn't give expected benefits even for
57 rotational disk drives and was becoming less relevant in the
58 age of SSDs. As none of the developers has been using
59 rotating media anymore, and nobody stepped up to actively
60 maintain this component of systemd it has now been removed.
62 * Swap units can use Discard= to specify discard options.
63 Discard options specified for swaps in /etc/fstab are now
66 * Docker containers are now detected as a separate type of
69 * The Password Agent protocol gained support for queries where
70 the user input is shown, useful e.g. for usernames.
71 systemd-ask-password gained a new --echo option to turn that
74 * The default sysctl.d/ snippets will now set:
76 net.core.default_qdisc = fq_codel
78 This selects Fair Queueing Controlled Delay as the default
79 queueing discipline for network interfaces. fq_codel helps
80 fight the network bufferbloat problem. It is believed to be
81 a good default with no tuning required for most workloads.
82 Downstream distributions may override this choice. On 10Gbit
83 servers that do not do forwarding, "fq" may perform better.
84 Systems without a good clocksource should use "pfifo_fast".
86 * If kdbus is enabled during build a new option BusPolicy= is
87 available for service units, that allows locking all service
88 processes into a stricter bus policy, in order to limit
89 access to various bus services, or even hide most of them
90 from the service's view entirely.
92 * networkctl will now show the .network and .link file
93 networkd has applied to a specific interface.
95 * sd-login gained a new API call sd_session_get_desktop() to
96 query which desktop environment has been selected for a
99 * UNIX utmp support is now compile-time optional to support
102 * systemctl gained two new commands "add-wants" and
103 "add-requires" for pulling in units from specific targets
106 * If the word "rescue" is specified on the kernel command line
107 the system will now boot into rescue mode (aka
108 rescue.target), which was previously available only by
109 specifying "1" or "systemd.unit=rescue.target" on the kernel
110 command line. This new kernel command line option nicely
111 mirrors the already existing "emergency" kernel command line
114 * New kernel command line options mount.usr=, mount.usrflags=,
115 usrfstype= have been added that match root=, rootflags=,
116 rootfstype= but allow mounting a specific file system to
119 * The $NOTIFY_SOCKET is now also passed to control processesof
120 services, not only the main process.
122 * This version reenables support for fsck's -l switch. This
123 means at least version v2.25 of util-linux is required for
124 operation, otherwise dead-locks on device nodes may
125 occur. Again: you need to update util-linux to at least
126 v2.25 when updating systemd to v217.
130 * timedated no longer reads NTP implementation unit names from
131 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
132 implementations should add a
134 Conflicts=systemd-timesyncd.service
136 to their unit files to take over and replace systemd's NTP
137 default functionality.
139 * systemd-sysusers gained a new line type "r" for configuring
140 which UID/GID ranges to allocate system users/groups
141 from. Lines of type "u" may now add an additional column
142 that specifies the home directory for the system user to be
143 created. Also, systemd-sysusers may now optionally read user
144 information from STDIN instead of a file. This is useful for
145 invoking it from RPM preinst scriptlets that need to create
146 users before the first RPM file is installed since these
147 files might need to be owned by them. A new
148 %sysusers_create_inline RPM macro has been introduced to do
149 just that. systemd-sysusers now updates the shadow files as
150 well as the user/group databases, which should enhance
151 compatibility with certain tools like grpck.
153 * A number of bus APIs of PID 1 now optionally consult
154 PolicyKit to permit access for otherwise unprivileged
155 clients under certain conditions. Note that this currently
156 doesn't support interactive authentication yet, but this is
157 expected to be added eventually, too.
159 * /etc/machine-info now has new fields for configuring the
160 deployment environment of the machine, as well as the
161 location of the machine. hostnamectl has been updated with
162 new command to update these fields.
164 * systemd-timesyncd has been updated to automatically acquire
165 NTP server information from systemd-networkd, which might
166 have been discovered via DHCP.
168 * systemd-resolved now includes a caching DNS stub resolver
169 and a complete LLMNR name resolution implementation. A new
170 NSS module "nss-resolve" has been added which can be used
171 instead of glibc's own "nss-dns" to resolve hostnames via
172 systemd-resolved. Hostnames, addresses and arbitrary RRs may
173 be resolved via systemd-resolved D-Bus APIs. In contrast to
174 the glibc internal resolver systemd-resolved is aware of
175 multi-homed system, and keeps DNS server and caches separate
176 and per-interface. Queries are sent simultaneously on all
177 interfaces that have DNS servers configured, in order to
178 properly handle VPNs and local LANs which might resolve
179 separate sets of domain names. systemd-resolved may acquire
180 DNS server information from systemd-networkd automatically,
181 which in turn might have discovered them via DHCP. A tool
182 "systemd-resolve-host" has been added that may be used to
183 query the DNS logic in resolved. systemd-resolved implements
184 IDNA and automatically uses IDNA or UTF-8 encoding depending
185 on whether classic DNS or LLMNR is used as transport. In the
186 next releases we intend to add a DNSSEC and mDNS/DNS-SD
187 implementation to systemd-resolved.
189 * A new NSS module nss-mymachines has been added, that
190 automatically resolves the names of all local registered
191 containers to their respective IP addresses.
193 * A new client tool "networkctl" for systemd-networkd has been
194 added. It currently is entirely passive and will query
195 networking configuration from udev, rtnetlink and networkd,
196 and present it to the user in a very friendly
197 way. Eventually, we hope to extend it to become a full
198 control utility for networkd.
200 * .socket units gained a new DeferAcceptSec= setting that
201 controls the kernels' TCP_DEFER_ACCEPT sockopt for
202 TCP. Similar, support for controlling TCP keep-alive
203 settings has been added (KeepAliveTimeSec=,
204 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
205 turning off Nagle's algorithm on TCP has been added
208 * logind learned a new session type "web", for use in projects
209 like Cockpit which register web clients as PAM sessions.
211 * timer units with at least one OnCalendar= setting will now
212 be started only after timer-sync.target has been
213 reached. This way they will not elapse before the system
214 clock has been corrected by a local NTP client or
215 similar. This is particular useful on RTC-less embedded
216 machines, that come up with an invalid system clock.
218 * systemd-nspawn's --network-veth= switch should now result in
219 stable MAC addresses for both the outer and the inner side
222 * systemd-nspawn gained a new --volatile= switch for running
223 container instances with /etc or /var unpopulated.
225 * The kdbus client code has been updated to use the new Linux
226 3.17 memfd subsystem instead of the old kdbus-specific one.
228 * systemd-networkd's DHCP client and server now support
229 FORCERENEW. There are also new configuration options to
230 configure the vendor client identifier and broadcast mode
233 * systemd will no longer inform the kernel about the current
234 timezone, as this is necessarily incorrect and racy as the
235 kernel has no understanding of DST and similar
236 concepts. This hence means FAT timestamps will be always
237 considered UTC, similar to what Android is already
238 doing. Also, when the RTC is configured to the local time
239 (rather than UTC) systemd will never synchronize back to it,
240 as this might confuse Windows at a later boot.
242 * systemd-analyze gained a new command "verify" for offline
243 validation of unit files.
245 * systemd-networkd gained support for a couple of additional
246 settings for bonding networking setups. Also, the metric for
247 statically configured routes may now be configured. For
248 network interfaces where this is appropriate the peer IP
249 address may now be configured.
251 * systemd-networkd's DHCP client will no longer request
252 broadcasting by default, as this tripped up some networks.
253 For hardware where broadcast is required the feature should
254 be switched back on using RequestBroadcast=yes.
256 * systemd-networkd will now set up IPv4LL addresses (when
257 enabled) even if DHCP is configured successfully.
259 * udev will now default to respect network device names given
260 by the kernel when the kernel indicates that these are
261 predictable. This behavior can be tweaked by changing
262 NamePolicy= in the relevant .link file.
264 * A new library systemd-terminal has been added that
265 implements full TTY stream parsing and rendering. This
266 library is supposed to be used later on for implementing a
267 full userspace VT subsystem, replacing the current kernel
270 * A new tool systemd-journal-upload has been added to push
271 journal data to a remote system running
272 systemd-journal-remote.
274 * journald will no longer forward all local data to another
275 running syslog daemon. This change has been made because
276 rsyslog (which appears to be the most commonly used syslog
277 implementation these days) no longer makes use of this, and
278 instead pulls the data out of the journal on its own. Since
279 forwarding the messages to a non-existent syslog server is
280 more expensive than we assumed we have now turned this
281 off. If you run a syslog server that is not a recent rsyslog
282 version, you have to turn this option on again
283 (ForwardToSyslog= in journald.conf).
285 * journald now optionally supports the LZ4 compressor for
286 larger journal fields. This compressor should perform much
287 better than XZ which was the previous default.
289 * machinectl now shows the IP addresses of local containers,
290 if it knows them, plus the interface name of the container.
292 * A new tool "systemd-escape" has been added that makes it
293 easy to escape strings to build unit names and similar.
295 * sd_notify() messages may now include a new ERRNO= field
296 which is parsed and collected by systemd and shown among the
297 "systemctl status" output for a service.
299 * A new component "systemd-firstboot" has been added that
300 queries the most basic systemd information (timezone,
301 hostname, root password) interactively on first
302 boot. Alternatively it may also be used to provision these
303 things offline on OS images installed into directories.
305 * The default sysctl.d/ snippets will now set
307 net.ipv4.conf.default.promote_secondaries=1
309 This has the benefit of no flushing secondary IP addresses
310 when primary addresses are removed.
312 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
313 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
314 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
315 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
316 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
317 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
318 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
319 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
320 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
321 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
322 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
323 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
324 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
325 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
326 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
328 -- Berlin, 2014-08-19
332 * A new tool systemd-sysusers has been added. This tool
333 creates system users and groups in /etc/passwd and
334 /etc/group, based on static declarative system user/group
335 definitions in /usr/lib/sysusers.d/. This is useful to
336 enable factory resets and volatile systems that boot up with
337 an empty /etc directory, and thus need system users and
338 groups created during early boot. systemd now also ships
339 with two default sysusers.d/ files for the most basic
340 users and groups systemd and the core operating system
343 * A new tmpfiles snippet has been added that rebuilds the
344 essential files in /etc on boot, should they be missing.
346 * A directive for ensuring automatic clean-up of
347 /var/cache/man/ has been removed from the default
348 configuration. This line should now be shipped by the man
349 implementation. The necessary change has been made to the
350 man-db implementation. Note that you need to update your man
351 implementation to one that ships this line, otherwise no
352 automatic clean-up of /var/cache/man will take place.
354 * A new condition ConditionNeedsUpdate= has been added that
355 may conditionalize services to only run when /etc or /var
356 are "older" than the vendor operating system resources in
357 /usr. This is useful for reconstructing or updating /etc
358 after an offline update of /usr or a factory reset, on the
359 next reboot. Services that want to run once after such an
360 update or reset should use this condition and order
361 themselves before the new systemd-update-done.service, which
362 will mark the two directories as fully updated. A number of
363 service files have been added making use of this, to rebuild
364 the udev hardware database, the journald message catalog and
365 dynamic loader cache (ldconfig). The systemd-sysusers tool
366 described above also makes use of this now. With this in
367 place it is now possible to start up a minimal operating
368 system with /etc empty cleanly. For more information on the
369 concepts involved see this recent blog story:
371 http://0pointer.de/blog/projects/stateless.html
373 * A new system group "input" has been introduced, and all
374 input device nodes get this group assigned. This is useful
375 for system-level software to get access to input devices. It
376 complements what is already done for "audio" and "video".
378 * systemd-networkd learnt minimal DHCPv4 server support in
379 addition to the existing DHCPv4 client support. It also
380 learnt DHCPv6 client and IPv6 Router Solicitation client
381 support. The DHCPv4 client gained support for static routes
382 passed in from the server. Note that the [DHCPv4] section
383 known in older systemd-networkd versions has been renamed to
384 [DHCP] and is now also used by the DHCPv6 client. Existing
385 .network files using settings of this section should be
386 updated, though compatibility is maintained. Optionally, the
387 client hostname may now be sent to the DHCP server.
389 * networkd gained support for vxlan virtual networks as well
390 as tun/tap and dummy devices.
392 * networkd gained support for automatic allocation of address
393 ranges for interfaces from a system-wide pool of
394 addresses. This is useful for dynamically managing a large
395 number of interfaces with a single network configuration
396 file. In particular this is useful to easily assign
397 appropriate IP addresses to the veth links of a large number
400 * RPM macros for processing sysusers, sysctl and binfmt
401 drop-in snippets at package installation time have been
404 * The /etc/os-release file should now be placed in
405 /usr/lib/os-release. The old location is automatically
406 created as symlink. /usr/lib is the more appropriate
407 location of this file, since it shall actually describe the
408 vendor operating system shipped in /usr, and not the
409 configuration stored in /etc.
411 * .mount units gained a new boolean SloppyOptions= setting
412 that maps to mount(8)'s -s option which enables permissive
413 parsing of unknown mount options.
415 * tmpfiles learnt a new "L+" directive which creates a symlink
416 but (unlike "L") deletes a pre-existing file first, should
417 it already exist and not already be the correct
418 symlink. Similar, "b+", "c+" and "p+" directives have been
419 added as well, which create block and character devices, as
420 well as fifos in the filesystem, possibly removing any
421 pre-existing files of different types.
423 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
424 'argument' field (which so far specified the source to
425 symlink/copy the files from) is now optional. If omitted the
426 same file os copied from /usr/share/factory/ suffixed by the
427 full destination path. This is useful for populating /etc
428 with essential files, by copying them from vendor defaults
429 shipped in /usr/share/factory/etc.
431 * A new command "systemctl preset-all" has been added that
432 applies the service preset settings to all installed unit
433 files. A new switch --preset-mode= has been added that
434 controls whether only enable or only disable operations
437 * A new command "systemctl is-system-running" has been added
438 that allows checking the overall state of the system, for
439 example whether it is fully up and running.
441 * When the system boots up with an empty /etc, the equivalent
442 to "systemctl preset-all" is executed during early boot, to
443 make sure all default services are enabled after a factory
446 * systemd now contains a minimal preset file that enables the
447 most basic services systemd ships by default.
449 * Unit files' [Install] section gained a new DefaultInstance=
450 field for defining the default instance to create if a
451 template unit is enabled with no instance specified.
453 * A new passive target cryptsetup-pre.target has been added
454 that may be used by services that need to make they run and
455 finish before the first LUKS cryptographic device is set up.
457 * The /dev/loop-control and /dev/btrfs-control device nodes
458 are now owned by the "disk" group by default, opening up
459 access to this group.
461 * systemd-coredump will now automatically generate a
462 stack trace of all core dumps taking place on the system,
463 based on elfutils' libdw library. This stack trace is logged
466 * systemd-coredump may now optionally store coredumps directly
467 on disk (in /var/lib/systemd/coredump, possibly compressed),
468 instead of storing them unconditionally in the journal. This
469 mode is the new default. A new configuration file
470 /etc/systemd/coredump.conf has been added to configure this
471 and other parameters of systemd-coredump.
473 * coredumpctl gained a new "info" verb to show details about a
474 specific coredump. A new switch "-1" has also been added
475 that makes sure to only show information about the most
476 recent entry instead of all entries. Also, as the tool is
477 generally useful now the "systemd-" prefix of the binary
478 name has been removed. Distributions that want to maintain
479 compatibility with the old name should add a symlink from
480 the old name to the new name.
482 * journald's SplitMode= now defaults to "uid". This makes sure
483 that unprivileged users can access their own coredumps with
484 coredumpctl without restrictions.
486 * New kernel command line options "systemd.wants=" (for
487 pulling an additional unit during boot), "systemd.mask="
488 (for masking a specific unit for the boot), and
489 "systemd.debug-shell" (for enabling the debug shell on tty9)
490 have been added. This is implemented in the new generator
491 "systemd-debug-generator".
493 * systemd-nspawn will now by default filter a couple of
494 syscalls for containers, among them those required for
495 kernel module loading, direct x86 IO port access, swap
496 management, and kexec. Most importantly though
497 open_by_handle_at() is now prohibited for containers,
498 closing a hole similar to a recently discussed vulnerability
499 in docker regarding access to files on file hierarchies the
500 container should normally not have access to. Note that for
501 nspawn we generally make no security claims anyway (and
502 this is explicitly documented in the man page), so this is
503 just a fix for one of the most obvious problems.
505 * A new man page file-hierarchy(7) has been added that
506 contains a minimized, modernized version of the file system
507 layout systemd expects, similar in style to the FHS
508 specification or hier(5). A new tool systemd-path(1) has
509 been added to query many of these paths for the local
512 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
513 longer done. Since the directory now has a per-user size
514 limit, and is cleaned on logout this appears unnecessary,
515 in particular since this now brings the lifecycle of this
516 directory closer in line with how IPC objects are handled.
518 * systemd.pc now exports a number of additional directories,
519 including $libdir (which is useful to identify the library
520 path for the primary architecture of the system), and a
521 couple of drop-in directories.
523 * udev's predictable network interface names now use the dev_port
524 sysfs attribute, introduced in linux 3.15 instead of dev_id to
525 distinguish between ports of the same PCI function. dev_id should
526 only be used for ports using the same HW address, hence the need
529 * machined has been updated to export the OS version of a
530 container (read from /etc/os-release and
531 /usr/lib/os-release) on the bus. This is now shown in
532 "machinectl status" for a machine.
534 * A new service setting RestartForceExitStatus= has been
535 added. If configured to a set of exit signals or process
536 return values, the service will be restarted when the main
537 daemon process exits with any of them, regardless of the
540 * systemctl's -H switch for connecting to remote systemd
541 machines has been extended so that it may be used to
542 directly connect to a specific container on the
543 host. "systemctl -H root@foobar:waldi" will now connect as
544 user "root" to host "foobar", and then proceed directly to
545 the container named "waldi". Note that currently you have to
546 authenticate as user "root" for this to work, as entering
547 containers is a privileged operation.
549 Contributions from: Andreas Henriksson, Benjamin Steinwender,
550 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
551 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
552 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
553 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
554 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
555 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
556 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
557 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
558 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
559 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
560 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
562 -- Berlin, 2014-07-03
566 * As an experimental feature, udev now tries to lock the
567 disk device node (flock(LOCK_SH|LOCK_NB)) while it
568 executes events for the disk or any of its partitions.
569 Applications like partitioning programs can lock the
570 disk device node (flock(LOCK_EX)) and claim temporary
571 device ownership that way; udev will entirely skip all event
572 handling for this disk and its partitions. If the disk
573 was opened for writing, the close will trigger a partition
574 table rescan in udev's "watch" facility, and if needed
575 synthesize "change" events for the disk and all its partitions.
576 This is now unconditionally enabled, and if it turns out to
577 cause major problems, we might turn it on only for specific
578 devices, or might need to disable it entirely. Device Mapper
579 devices are excluded from this logic.
581 * We temporarily dropped the "-l" switch for fsck invocations,
582 since they collide with the flock() logic above. util-linux
583 upstream has been changed already to avoid this conflict,
584 and we will readd "-l" as soon as util-linux with this
585 change has been released.
587 * The dependency on libattr has been removed. Since a long
588 time, the extended attribute calls have moved to glibc, and
589 libattr is thus unnecessary.
591 * Virtualization detection works without priviliges now. This
592 means the systemd-detect-virt binary no longer requires
593 CAP_SYS_PTRACE file capabilities, and our daemons can run
594 with fewer privileges.
596 * systemd-networkd now runs under its own "systemd-network"
597 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
598 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
599 loses the ability to write to files owned by root this way.
601 * Similar, systemd-resolved now runs under its own
602 "systemd-resolve" user with no capabilities remaining.
604 * Similar, systemd-bus-proxyd now runs under its own
605 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
607 * systemd-networkd gained support for setting up "veth"
608 virtual ethernet devices for container connectivity, as well
609 as GRE and VTI tunnels.
611 * systemd-networkd will no longer automatically attempt to
612 manually load kernel modules necessary for certain tunnel
613 transports. Instead, it is assumed the kernel loads them
614 automatically when required. This only works correctly on
615 very new kernels. On older kernels, please consider adding
616 the kernel modules to /etc/modules-load.d/ as a work-around.
618 * The resolv.conf file systemd-resolved generates has been
619 moved to /run/systemd/resolve/. If you have a symlink from
620 /etc/resolv.conf, it might be necessary to correct it.
622 * Two new service settings, ProtectHome= and ProtectSystem=,
623 have been added. When enabled, they will make the user data
624 (such as /home) inaccessible or read-only and the system
625 (such as /usr) read-only, for specific services. This allows
626 very light-weight per-service sandboxing to avoid
627 modifications of user data or system files from
628 services. These two new switches have been enabled for all
629 of systemd's long-running services, where appropriate.
631 * Socket units gained new SocketUser= and SocketGroup=
632 settings to set the owner user and group of AF_UNIX sockets
633 and FIFOs in the file system.
635 * Socket units gained a new RemoveOnStop= setting. If enabled,
636 all FIFOS and sockets in the file system will be removed
637 when the specific socket unit is stopped.
639 * Socket units gained a new Symlinks= setting. It takes a list
640 of symlinks to create to file system sockets or FIFOs
641 created by the specific Unix sockets. This is useful to
642 manage symlinks to socket nodes with the same life-cycle as
645 * The /dev/log socket and /dev/initctl FIFO have been moved to
646 /run, and have been replaced by symlinks. This allows
647 connecting to these facilities even if PrivateDevices=yes is
648 used for a service (which makes /dev/log itself unavailable,
649 but /run is left). This also has the benefit of ensuring
650 that /dev only contains device nodes, directories and
651 symlinks, and nothing else.
653 * sd-daemon gained two new calls sd_pid_notify() and
654 sd_pid_notifyf(). They are similar to sd_notify() and
655 sd_notifyf(), but allow overriding of the source PID of
656 notification messages if permissions permit this. This is
657 useful to send notify messages on behalf of a different
658 process (for example, the parent process). The
659 systemd-notify tool has been updated to make use of this
660 when sending messages (so that notification messages now
661 originate from the shell script invoking systemd-notify and
662 not the systemd-notify process itself. This should minimize
663 a race where systemd fails to associate notification
664 messages to services when the originating process already
667 * A new "on-abnormal" setting for Restart= has been added. If
668 set, it will result in automatic restarts on all "abnormal"
669 reasons for a process to exit, which includes unclean
670 signals, core dumps, timeouts and watchdog timeouts, but
671 does not include clean and unclean exit codes or clean
672 signals. Restart=on-abnormal is an alternative for
673 Restart=on-failure for services that shall be able to
674 terminate and avoid restarts on certain errors, by
675 indicating so with an unclean exit code. Restart=on-failure
676 or Restart=on-abnormal is now the recommended setting for
677 all long-running services.
679 * If the InaccessibleDirectories= service setting points to a
680 mount point (or if there are any submounts contained within
681 it), it is now attempted to completely unmount it, to make
682 the file systems truly unavailable for the respective
685 * The ReadOnlyDirectories= service setting and
686 systemd-nspawn's --read-only parameter are now recursively
687 applied to all submounts, too.
689 * Mount units may now be created transiently via the bus APIs.
691 * The support for SysV and LSB init scripts has been removed
692 from the systemd daemon itself. Instead, it is now
693 implemented as a generator that creates native systemd units
694 from these scripts when needed. This enables us to remove a
695 substantial amount of legacy code from PID 1, following the
696 fact that many distributions only ship a very small number
697 of LSB/SysV init scripts nowadays.
699 * Priviliged Xen (dom0) domains are not considered
700 virtualization anymore by the virtualization detection
701 logic. After all, they generally have unrestricted access to
702 the hardware and usually are used to manage the unprivileged
705 * systemd-tmpfiles gained a new "C" line type, for copying
706 files or entire directories.
708 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
709 lines. So far, they have been non-globbing versions of the
710 latter, and have thus been redundant. In future, it is
711 recommended to only use "z". "m" has hence been removed
712 from the documentation, even though it stays supported.
714 * A tmpfiles snippet to recreate the most basic structure in
715 /var has been added. This is enough to create the /var/run →
716 /run symlink and create a couple of structural
717 directories. This allows systems to boot up with an empty or
718 volatile /var. Of course, while with this change, the core OS
719 now is capable with dealing with a volatile /var, not all
720 user services are ready for it. However, we hope that sooner
721 or later, many service daemons will be changed upstream so
722 that they are able to automatically create their necessary
723 directories in /var at boot, should they be missing. This is
724 the first step to allow state-less systems that only require
725 the vendor image for /usr to boot.
727 * systemd-nspawn has gained a new --tmpfs= switch to mount an
728 empty tmpfs instance to a specific directory. This is
729 particularly useful for making use of the automatic
730 reconstruction of /var (see above), by passing --tmpfs=/var.
732 * Access modes specified in tmpfiles snippets may now be
733 prefixed with "~", which indicates that they shall be masked
734 by whether the existing file or directory is currently
735 writable, readable or executable at all. Also, if specified,
736 the sgid/suid/sticky bits will be masked for all
739 * A new passive target unit "network-pre.target" has been
740 added which is useful for services that shall run before any
741 network is configured, for example firewall scripts.
743 * The "floppy" group that previously owned the /dev/fd*
744 devices is no longer used. The "disk" group is now used
745 instead. Distributions should probably deprecate usage of
748 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
749 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
750 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
751 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
752 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
753 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
756 -- Berlin, 2014-06-11
760 * A new "systemd-timesyncd" daemon has been added for
761 synchronizing the system clock across the network. It
762 implements an SNTP client. In contrast to NTP
763 implementations such as chrony or the NTP reference server,
764 this only implements a client side, and does not bother with
765 the full NTP complexity, focusing only on querying time from
766 one remote server and synchronizing the local clock to
767 it. Unless you intend to serve NTP to networked clients or
768 want to connect to local hardware clocks, this simple NTP
769 client should be more than appropriate for most
770 installations. The daemon runs with minimal privileges, and
771 has been hooked up with networkd to only operate when
772 network connectivity is available. The daemon saves the
773 current clock to disk every time a new NTP sync has been
774 acquired, and uses this to possibly correct the system clock
775 early at bootup, in order to accommodate for systems that
776 lack an RTC such as the Raspberry Pi and embedded devices,
777 and to make sure that time monotonically progresses on these
778 systems, even if it is not always correct. To make use of
779 this daemon, a new system user and group "systemd-timesync"
780 needs to be created on installation of systemd.
782 * The queue "seqnum" interface of libudev has been disabled, as
783 it was generally incompatible with device namespacing as
784 sequence numbers of devices go "missing" if the devices are
785 part of a different namespace.
787 * "systemctl list-timers" and "systemctl list-sockets" gained
788 a --recursive switch for showing units of these types also
789 for all local containers, similar in style to the already
790 supported --recursive switch for "systemctl list-units".
792 * A new RebootArgument= setting has been added for service
793 units, which may be used to specify a kernel reboot argument
794 to use when triggering reboots with StartLimitAction=.
796 * A new FailureAction= setting has been added for service
797 units which may be used to specify an operation to trigger
798 when a service fails. This works similarly to
799 StartLimitAction=, but unlike it, controls what is done
800 immediately rather than only after several attempts to
801 restart the service in question.
803 * hostnamed got updated to also expose the kernel name,
804 release, and version on the bus. This is useful for
805 executing commands like hostnamectl with the -H switch.
806 systemd-analyze makes use of this to properly display
807 details when running non-locally.
809 * The bootchart tool can now show cgroup information in the
812 * The CFS CPU quota cgroup attribute is now exposed for
813 services. The new CPUQuota= switch has been added for this
814 which takes a percentage value. Setting this will have the
815 result that a service may never get more CPU time than the
816 specified percentage, even if the machine is otherwise idle.
818 * systemd-networkd learned IPIP and SIT tunnel support.
820 * LSB init scripts exposing a dependency on $network will now
821 get a dependency on network-online.target rather than simply
822 network.target. This should bring LSB handling closer to
823 what it was on SysV systems.
825 * A new fsck.repair= kernel option has been added to control
826 how fsck shall deal with unclean file systems at boot.
828 * The (.ini) configuration file parser will now silently
829 ignore sections whose name begins with "X-". This may be
830 used to maintain application-specific extension sections in unit
833 * machined gained a new API to query the IP addresses of
834 registered containers. "machinectl status" has been updated
835 to show these addresses in its output.
837 * A new call sd_uid_get_display() has been added to the
838 sd-login APIs for querying the "primary" session of a
839 user. The "primary" session of the user is elected from the
840 user's sessions and generally a graphical session is
841 preferred over a text one.
843 * A minimal systemd-resolved daemon has been added. It
844 currently simply acts as a companion to systemd-networkd and
845 manages resolv.conf based on per-interface DNS
846 configuration, possibly supplied via DHCP. In the long run
847 we hope to extend this into a local DNSSEC enabled DNS and
850 * The systemd-networkd-wait-online tool is now enabled by
851 default. It will delay network-online.target until a network
852 connection has been configured. The tool primarily integrates
853 with networkd, but will also make a best effort to make sense
854 of network configuration performed in some other way.
856 * Two new service options StartupCPUShares= and
857 StartupBlockIOWeight= have been added that work similarly to
858 CPUShares= and BlockIOWeight= however only apply during
859 system startup. This is useful to prioritize certain services
860 differently during bootup than during normal runtime.
862 * hostnamed has been changed to prefer the statically
863 configured hostname in /etc/hostname (unless set to
864 'localhost' or empty) over any dynamic one supplied by
865 dhcp. With this change, the rules for picking the hostname
866 match more closely the rules of other configuration settings
867 where the local administrator's configuration in /etc always
868 overrides any other settings.
870 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
871 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
872 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
873 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
874 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
875 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
876 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
877 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
878 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
879 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
880 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
881 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
882 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
883 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
884 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
885 Lindskog, WaLyong Cho, Will Woods, Zbigniew
888 -- Beijing, 2014-05-28
892 * When restoring the screen brightness at boot, stay away from
893 the darkest setting or from the lowest 5% of the available
894 range, depending on which is the larger value of both. This
895 should effectively protect the user from rebooting into a
896 black screen, should the brightness have been set to minimum
899 * sd-login gained a new sd_machine_get_class() call to
900 determine the class ("vm" or "container") of a machine
901 registered with machined.
903 * sd-login gained new calls
904 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
905 to query the identity of the peer of a local AF_UNIX
906 connection. They operate similarly to their sd_pid_get_xyz()
909 * PID 1 will now maintain a system-wide system state engine
910 with the states "starting", "running", "degraded",
911 "maintenance", "stopping". These states are bound to system
912 startup, normal runtime, runtime with at least one failed
913 service, rescue/emergency mode and system shutdown. This
914 state is shown in the "systemctl status" output when no unit
915 name is passed. It is useful to determine system state, in
916 particularly when doing so for many systems or containers at
919 * A new command "list-machines" has been added to "systemctl"
920 that lists all local OS containers and shows their system
921 state (see above), if systemd runs inside of them.
923 * systemctl gained a new "-r" switch to recursively enumerate
924 units on all local containers, when used with the
925 "list-unit" command (which is the default one that is
926 executed when no parameters are specified).
928 * The GPT automatic partition discovery logic will now honour
929 two GPT partition flags: one may be set on a partition to
930 cause it to be mounted read-only, and the other may be set
931 on a partition to ignore it during automatic discovery.
933 * Two new GPT type UUIDs have been added for automatic root
934 partition discovery, for 32-bit and 64-bit ARM. This is not
935 particularly useful for discovering the root directory on
936 these architectures during bare-metal boots (since UEFI is
937 not common there), but still very useful to allow booting of
938 ARM disk images in nspawn with the -i option.
940 * MAC addresses of interfaces created with nspawn's
941 --network-interface= switch will now be generated from the
942 machine name, and thus be stable between multiple invocations
945 * logind will now automatically remove all IPC objects owned
946 by a user if she or he fully logs out. This makes sure that
947 users who are logged out cannot continue to consume IPC
948 resources. This covers SysV memory, semaphores and message
949 queues as well as POSIX shared memory and message
950 queues. Traditionally, SysV and POSIX IPC had no life-cycle
951 limits. With this functionality, that is corrected. This may
952 be turned off by using the RemoveIPC= switch of logind.conf.
954 * The systemd-machine-id-setup and tmpfiles tools gained a
955 --root= switch to operate on a specific root directory,
958 * journald can now forward logged messages to the TTYs of all
959 logged in users ("wall"). This is the default for all
960 emergency messages now.
962 * A new tool systemd-journal-remote has been added to stream
963 journal log messages across the network.
965 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
966 controller trees are mounted into it. Note that the
967 directories mounted beneath it are not read-only. This is a
968 security measure and is particularly useful because glibc
969 actually includes a search logic to pick any tmpfs it can
970 find to implement shm_open() if /dev/shm is not available
971 (which it might very well be in namespaced setups).
973 * machinectl gained a new "poweroff" command to cleanly power
974 down a local OS container.
976 * The PrivateDevices= unit file setting will now also drop the
977 CAP_MKNOD capability from the capability bound set, and
978 imply DevicePolicy=closed.
980 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
981 comprehensively on all long-running systemd services where
984 * systemd-udevd will now run in a disassociated mount
985 namespace. To mount directories from udev rules, make sure to
986 pull in mount units via SYSTEMD_WANTS properties.
988 * The kdbus support gained support for uploading policy into
989 the kernel. sd-bus gained support for creating "monitoring"
990 connections that can eavesdrop into all bus communication
991 for debugging purposes.
993 * Timestamps may now be specified in seconds since the UNIX
994 epoch Jan 1st, 1970 by specifying "@" followed by the value
997 * Native tcpwrap support in systemd has been removed. tcpwrap
998 is old code, not really maintained anymore and has serious
999 shortcomings, and better options such as firewalls
1000 exist. For setups that require tcpwrap usage, please
1001 consider invoking your socket-activated service via tcpd,
1002 like on traditional inetd.
1004 * A new system.conf configuration option
1005 DefaultTimerAccuracySec= has been added that controls the
1006 default AccuracySec= setting of .timer units.
1008 * Timer units gained a new WakeSystem= switch. If enabled,
1009 timers configured this way will cause the system to resume
1010 from system suspend (if the system supports that, which most
1013 * Timer units gained a new Persistent= switch. If enabled,
1014 timers configured this way will save to disk when they have
1015 been last triggered. This information is then used on next
1016 reboot to possible execute overdue timer events, that
1017 could not take place because the system was powered off.
1018 This enables simple anacron-like behaviour for timer units.
1020 * systemctl's "list-timers" will now also list the time a
1021 timer unit was last triggered in addition to the next time
1022 it will be triggered.
1024 * systemd-networkd will now assign predictable IPv4LL
1025 addresses to its local interfaces.
1027 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
1028 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
1029 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
1030 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
1031 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
1032 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
1033 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
1034 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
1037 -- Berlin, 2014-03-25
1041 * A new unit file setting RestrictAddressFamilies= has been
1042 added to restrict which socket address families unit
1043 processes gain access to. This takes address family names
1044 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1045 attack surface of services via exotic protocol stacks. This
1046 is built on seccomp system call filters.
1048 * Two new unit file settings RuntimeDirectory= and
1049 RuntimeDirectoryMode= have been added that may be used to
1050 manage a per-daemon runtime directories below /run. This is
1051 an alternative for setting up directory permissions with
1052 tmpfiles snippets, and has the advantage that the runtime
1053 directory's lifetime is bound to the daemon runtime and that
1054 the daemon starts up with an empty directory each time. This
1055 is particularly useful when writing services that drop
1056 privileges using the User= or Group= setting.
1058 * The DeviceAllow= unit setting now supports globbing for
1059 matching against device group names.
1061 * The systemd configuration file system.conf gained new
1062 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1063 DefaultMemoryAccounting= to globally turn on/off accounting
1064 for specific resources (cgroups) for all units. These
1065 settings may still be overridden individually in each unit
1068 * systemd-gpt-auto-generator is now able to discover /srv and
1069 root partitions in addition to /home and swap partitions. It
1070 also supports LUKS-encrypted partitions now. With this in
1071 place, automatic discovery of partitions to mount following
1072 the Discoverable Partitions Specification
1073 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1074 is now a lot more complete. This allows booting without
1075 /etc/fstab and without root= on the kernel command line on
1076 systems prepared appropriately.
1078 * systemd-nspawn gained a new --image= switch which allows
1079 booting up disk images and Linux installations on any block
1080 device that follow the Discoverable Partitions Specification
1081 (see above). This means that installations made with
1082 appropriately updated installers may now be started and
1083 deployed using container managers, completely
1084 unmodified. (We hope that libvirt-lxc will add support for
1085 this feature soon, too.)
1087 * systemd-nspawn gained a new --network-macvlan= setting to
1088 set up a private macvlan interface for the
1089 container. Similarly, systemd-networkd gained a new
1090 Kind=macvlan setting in .netdev files.
1092 * systemd-networkd now supports configuring local addresses
1095 * A new tool systemd-network-wait-online has been added to
1096 synchronously wait for network connectivity using
1099 * The sd-bus.h bus API gained a new sd_bus_track object for
1100 tracking the life-cycle of bus peers. Note that sd-bus.h is
1101 still not a public API though (unless you specify
1102 --enable-kdbus on the configure command line, which however
1103 voids your warranty and you get no API stability guarantee).
1105 * The $XDG_RUNTIME_DIR runtime directories for each user are
1106 now individual tmpfs instances, which has the benefit of
1107 introducing separate pools for each user, with individual
1108 size limits, and thus making sure that unprivileged clients
1109 can no longer negatively impact the system or other users by
1110 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
1111 RuntimeDirectorySize= has been introduced that allows
1112 controlling the default size limit for all users. It
1113 defaults to 10% of the available physical memory. This is no
1114 replacement for quotas on tmpfs though (which the kernel
1115 still does not support), as /dev/shm and /tmp are still
1116 shared resources used by both the system and unprivileged
1119 * logind will now automatically turn off automatic suspending
1120 on laptop lid close when more than one display is
1121 connected. This was previously expected to be implemented
1122 individually in desktop environments (such as GNOME),
1123 however has been added to logind now, in order to fix a
1124 boot-time race where a desktop environment might not have
1125 been started yet and thus not been able to take an inhibitor
1126 lock at the time where logind already suspends the system
1127 due to a closed lid.
1129 * logind will now wait at least 30s after each system
1130 suspend/resume cycle, and 3min after system boot before
1131 suspending the system due to a closed laptop lid. This
1132 should give USB docking stations and similar enough time to
1133 be probed and configured after system resume and boot in
1134 order to then act as suspend blocker.
1136 * systemd-run gained a new --property= setting which allows
1137 initialization of resource control properties (and others)
1138 for the created scope or service unit. Example: "systemd-run
1139 --property=BlockIOWeight=10 updatedb" may be used to run
1140 updatedb at a low block IO scheduling weight.
1142 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1143 now also work in --scope mode.
1145 * When systemd is compiled with kdbus support, basic support
1146 for enforced policies is now in place. (Note that enabling
1147 kdbus still voids your warranty and no API compatibility
1150 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1151 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1152 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1153 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1154 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1155 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1156 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1157 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
1158 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
1159 Zbigniew Jędrzejewski-Szmek
1161 -- Berlin, 2014-03-12
1165 * systemd will now relabel /dev after loading the SMACK policy
1166 according to SMACK rules.
1168 * A new unit file option AppArmorProfile= has been added to
1169 set the AppArmor profile for the processes of a unit.
1171 * A new condition check ConditionArchitecture= has been added
1172 to conditionalize units based on the system architecture, as
1173 reported by uname()'s "machine" field.
1175 * systemd-networkd now supports matching on the system
1176 virtualization, architecture, kernel command line, host name
1179 * logind is now a lot more aggressive when suspending the
1180 machine due to a closed laptop lid. Instead of acting only
1181 on the lid close action, it will continuously watch the lid
1182 status and act on it. This is useful for laptops where the
1183 power button is on the outside of the chassis so that it can
1184 be reached without opening the lid (such as the Lenovo
1185 Yoga). On those machines, logind will now immediately
1186 re-suspend the machine if the power button has been
1187 accidentally pressed while the laptop was suspended and in a
1188 backpack or similar.
1190 * logind will now watch SW_DOCK switches and inhibit reaction
1191 to the lid switch if it is pressed. This means that logind
1192 will not suspend the machine anymore if the lid is closed
1193 and the system is docked, if the laptop supports SW_DOCK
1194 notifications via the input layer. Note that ACPI docking
1195 stations do not generate this currently. Also note that this
1196 logic is usually not fully sufficient and Desktop
1197 Environments should take a lid switch inhibitor lock when an
1198 external display is connected, as systemd will not watch
1201 * nspawn will now make use of the devices cgroup controller by
1202 default, and only permit creation of and access to the usual
1203 API device nodes like /dev/null or /dev/random, as well as
1204 access to (but not creation of) the pty devices.
1206 * We will now ship a default .network file for
1207 systemd-networkd that automatically configures DHCP for
1208 network interfaces created by nspawn's --network-veth or
1209 --network-bridge= switches.
1211 * systemd will now understand the usual M, K, G, T suffixes
1212 according to SI conventions (i.e. to the base 1000) when
1213 referring to throughput and hardware metrics. It will stay
1214 with IEC conventions (i.e. to the base 1024) for software
1215 metrics, according to what is customary according to
1216 Wikipedia. We explicitly document which base applies for
1217 each configuration option.
1219 * The DeviceAllow= setting in unit files now supports a syntax
1220 to whitelist an entire group of devices node majors at once,
1221 based on the /proc/devices listing. For example, with the
1222 string "char-pts", it is now possible to whitelist all
1223 current and future pseudo-TTYs at once.
1225 * sd-event learned a new "post" event source. Event sources of
1226 this type are triggered by the dispatching of any event
1227 source of a type that is not "post". This is useful for
1228 implementing clean-up and check event sources that are
1229 triggered by other work being done in the program.
1231 * systemd-networkd is no longer statically enabled, but uses
1232 the usual [Install] sections so that it can be
1233 enabled/disabled using systemctl. It still is enabled by
1236 * When creating a veth interface pair with systemd-nspawn, the
1237 host side will now be prefixed with "vb-" if
1238 --network-bridge= is used, and with "ve-" if --network-veth
1239 is used. This way, it is easy to distinguish these cases on
1240 the host, for example to apply different configuration to
1241 them with systemd-networkd.
1243 * The compatibility libraries for libsystemd-journal.so,
1244 libsystem-id128.so, libsystemd-login.so and
1245 libsystemd-daemon.so do not make use of IFUNC
1246 anymore. Instead, we now build libsystemd.so multiple times
1247 under these alternative names. This means that the footprint
1248 is drastically increased, but given that these are
1249 transitional compatibility libraries, this should not matter
1250 much. This change has been made necessary to support the ARM
1251 platform for these compatibility libraries, as the ARM
1252 toolchain is not really at the same level as the toolchain
1253 for other architectures like x86 and does not support
1254 IFUNC. Please make sure to use --enable-compat-libs only
1255 during a transitional period!
1257 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
1258 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1259 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
1260 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
1261 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
1262 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
1263 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
1264 Zbigniew Jędrzejewski-Szmek
1266 -- Berlin, 2014-02-24
1270 * A new component "systemd-networkd" has been added that can
1271 be used to configure local network interfaces statically or
1272 via DHCP. It is capable of bringing up bridges, VLANs, and
1273 bonding. Currently, no hook-ups for interactive network
1274 configuration are provided. Use this for your initrd,
1275 container, embedded, or server setup if you need a simple,
1276 yet powerful, network configuration solution. This
1277 configuration subsystem is quite nifty, as it allows wildcard
1278 hotplug matching in interfaces. For example, with a single
1279 configuration snippet, you can configure that all Ethernet
1280 interfaces showing up are automatically added to a bridge,
1281 or similar. It supports link-sensing and more.
1283 * A new tool "systemd-socket-proxyd" has been added which can
1284 act as a bidirectional proxy for TCP sockets. This is
1285 useful for adding socket activation support to services that
1286 do not actually support socket activation, including virtual
1287 machines and the like.
1289 * Add a new tool to save/restore rfkill state on
1292 * Save/restore state of keyboard backlights in addition to
1293 display backlights on shutdown/boot.
1295 * udev learned a new SECLABEL{} construct to label device
1296 nodes with a specific security label when they appear. For
1297 now, only SECLABEL{selinux} is supported, but the syntax is
1298 prepared for additional security frameworks.
1300 * udev gained a new scheme to configure link-level attributes
1301 from files in /etc/systemd/network/*.link. These files can
1302 match against MAC address, device path, driver name and type,
1303 and will apply attributes like the naming policy, link speed,
1304 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
1305 address assignment policy (randomized, ...).
1307 * The configuration of network interface naming rules for
1308 "permanent interface names" has changed: a new NamePolicy=
1309 setting in the [Link] section of .link files determines the
1310 priority of possible naming schemes (onboard, slot, mac,
1311 path). The default value of this setting is determined by
1312 /usr/lib/net/links/99-default.link. Old
1313 80-net-name-slot.rules udev configuration file has been
1314 removed, so local configuration overriding this file should
1315 be adapated to override 99-default.link instead.
1317 * When the User= switch is used in a unit file, also
1318 initialize $SHELL= based on the user database entry.
1320 * systemd no longer depends on libdbus. All communication is
1321 now done with sd-bus, systemd's low-level bus library
1324 * kdbus support has been added to PID 1 itself. When kdbus is
1325 enabled, this causes PID 1 to set up the system bus and
1326 enable support for a new ".busname" unit type that
1327 encapsulates bus name activation on kdbus. It works a little
1328 bit like ".socket" units, except for bus names. A new
1329 generator has been added that converts classic dbus1 service
1330 activation files automatically into native systemd .busname
1333 * sd-bus: add a light-weight vtable implementation that allows
1334 defining objects on the bus with a simple static const
1335 vtable array of its methods, signals and properties.
1337 * systemd will not generate or install static dbus
1338 introspection data anymore to /usr/share/dbus-1/interfaces,
1339 as the precise format of these files is unclear, and
1340 nothing makes use of it.
1342 * A proxy daemon is now provided to proxy clients connecting
1343 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1344 compatibility with classic D-Bus.
1346 * A bus driver implementation has been added that supports the
1347 classic D-Bus bus driver calls on kdbus, also for
1348 compatibility purposes.
1350 * A new API "sd-event.h" has been added that implements a
1351 minimal event loop API built around epoll. It provides a
1352 couple of features that direct epoll usage is lacking:
1353 prioritization of events, scales to large numbers of timer
1354 events, per-event timer slack (accuracy), system-wide
1355 coalescing of timer events, exit handlers, watchdog
1356 supervision support using systemd's sd_notify() API, child
1359 * A new API "sd-rntl.h" has been added that provides an API
1360 around the route netlink interface of the kernel, similar in
1361 style to "sd-bus.h".
1363 * A new API "sd-dhcp-client.h" has been added that provides a
1364 small DHCPv4 client-side implementation. This is used by
1367 * There is a new kernel command line option
1368 "systemd.restore_state=0|1". When set to "0", none of the
1369 systemd tools will restore saved runtime state to hardware
1370 devices. More specifically, the rfkill and backlight states
1373 * The FsckPassNo= compatibility option in mount/service units
1374 has been removed. The fstab generator will now add the
1375 necessary dependencies automatically, and does not require
1376 PID1's support for that anymore.
1378 * journalctl gained a new switch, --list-boots, that lists
1379 recent boots with their times and boot IDs.
1381 * The various tools like systemctl, loginctl, timedatectl,
1382 busctl, systemd-run, ... have gained a new switch "-M" to
1383 connect to a specific, local OS container (as direct
1384 connection, without requiring SSH). This works on any
1385 container that is registered with machined, such as those
1386 created by libvirt-lxc or nspawn.
1388 * systemd-run and systemd-analyze also gained support for "-H"
1389 to connect to remote hosts via SSH. This is particularly
1390 useful for systemd-run because it enables queuing of jobs
1391 onto remote systems.
1393 * machinectl gained a new command "login" to open a getty
1394 login in any local container. This works with any container
1395 that is registered with machined (such as those created by
1396 libvirt-lxc or nspawn), and which runs systemd inside.
1398 * machinectl gained a new "reboot" command that may be used to
1399 trigger a reboot on a specific container that is registered
1400 with machined. This works on any container that runs an init
1401 system of some kind.
1403 * systemctl gained a new "list-timers" command to print a nice
1404 listing of installed timer units with the times they elapse
1407 * Alternative reboot() parameters may now be specified on the
1408 "systemctl reboot" command line and are passed to the
1409 reboot() system call.
1411 * systemctl gained a new --job-mode= switch to configure the
1412 mode to queue a job with. This is a more generic version of
1413 --fail, --irreversible, and --ignore-dependencies, which are
1414 still available but not advertised anymore.
1416 * /etc/systemd/system.conf gained new settings to configure
1417 various default timeouts of units, as well as the default
1418 start limit interval and burst. These may still be overridden
1421 * PID1 will now export on the bus profile data of the security
1422 policy upload process (such as the SELinux policy upload to
1425 * journald: when forwarding logs to the console, include
1426 timestamps (following the setting in
1427 /sys/module/printk/parameters/time).
1429 * OnCalendar= in timer units now understands the special
1430 strings "yearly" and "annually". (Both are equivalent)
1432 * The accuracy of timer units is now configurable with the new
1433 AccuracySec= setting. It defaults to 1min.
1435 * A new dependency type JoinsNamespaceOf= has been added that
1436 allows running two services within the same /tmp and network
1437 namespace, if PrivateNetwork= or PrivateTmp= are used.
1439 * A new command "cat" has been added to systemctl. It outputs
1440 the original unit file of a unit, and concatenates the
1441 contents of additional "drop-in" unit file snippets, so that
1442 the full configuration is shown.
1444 * systemctl now supports globbing on the various "list-xyz"
1445 commands, like "list-units" or "list-sockets", as well as on
1446 those commands which take multiple unit names.
1448 * journalctl's --unit= switch gained support for globbing.
1450 * All systemd daemons now make use of the watchdog logic so
1451 that systemd automatically notices when they hang.
1453 * If the $container_ttys environment variable is set,
1454 getty-generator will automatically spawn a getty for each
1455 listed tty. This is useful for container managers to request
1456 login gettys to be spawned on as many ttys as needed.
1458 * %h, %s, %U specifier support is not available anymore when
1459 used in unit files for PID 1. This is because NSS calls are
1460 not safe from PID 1. They stay available for --user
1461 instances of systemd, and as special case for the root user.
1463 * loginctl gained a new "--no-legend" switch to turn off output
1466 * The "sd-login.h" API gained three new calls:
1467 sd_session_is_remote(), sd_session_get_remote_user(),
1468 sd_session_get_remote_host() to query information about
1471 * The udev hardware database now also carries vendor/product
1472 information of SDIO devices.
1474 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
1475 determine whether watchdog notifications are requested by
1478 * Socket-activated per-connection services now include a
1479 short description of the connection parameters in the
1482 * tmpfiles gained a new "--boot" option. When this is not used,
1483 only lines where the command character is not suffixed with
1484 "!" are executed. When this option is specified, those
1485 options are executed too. This partitions tmpfiles
1486 directives into those that can be safely executed at any
1487 time, and those which should be run only at boot (for
1488 example, a line that creates /run/nologin).
1490 * A new API "sd-resolve.h" has been added which provides a simple
1491 asynchronous wrapper around glibc NSS host name resolution
1492 calls, such as getaddrinfo(). In contrast to glibc's
1493 getaddrinfo_a(), it does not use signals. In contrast to most
1494 other asynchronous name resolution libraries, this one does
1495 not reimplement DNS, but reuses NSS, so that alternate
1496 host name resolution systems continue to work, such as mDNS,
1497 LDAP, etc. This API is based on libasyncns, but it has been
1498 cleaned up for inclusion in systemd.
1500 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
1501 "sd-daemon.h" are no longer found in individual libraries
1502 libsystemd-journal.so, libsystemd-login.so,
1503 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
1504 merged them into a single library, libsystemd.so, which
1505 provides all symbols. The reason for this is cyclic
1506 dependencies, as these libraries tend to use each other's
1507 symbols. So far, we have managed to workaround that by linking
1508 a copy of a good part of our code into each of these
1509 libraries again and again, which, however, makes certain
1510 things hard to do, like sharing static variables. Also, it
1511 substantially increases footprint. With this change, there
1512 is only one library for the basic APIs systemd
1513 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
1514 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
1515 library as well, however are subject to the --enable-kdbus
1516 switch (see below). Note that "sd-dhcp-client.h" is not part
1517 of this library (this is because it only consumes, never
1518 provides, services of/to other APIs). To make the transition
1519 easy from the separate libraries to the unified one, we
1520 provide the --enable-compat-libs compile-time switch which
1521 will generate stub libraries that are compatible with the
1522 old ones but redirect all calls to the new one.
1524 * All of the kdbus logic and the new APIs "sd-bus.h",
1525 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
1526 and "sd-utf8.h" are compile-time optional via the
1527 "--enable-kdbus" switch, and they are not compiled in by
1528 default. To make use of kdbus, you have to explicitly enable
1529 the switch. Note however, that neither the kernel nor the
1530 userspace API for all of this is considered stable yet. We
1531 want to maintain the freedom to still change the APIs for
1532 now. By specifying this build-time switch, you acknowledge
1533 that you are aware of the instability of the current
1536 * Also, note that while kdbus is pretty much complete,
1537 it lacks one thing: proper policy support. This means you
1538 can build a fully working system with all features; however,
1539 it will be highly insecure. Policy support will be added in
1540 one of the next releases, at the same time that we will
1541 declare the APIs stable.
1543 * When the kernel command-line argument "kdbus" is specified,
1544 systemd will automatically load the kdbus.ko kernel module. At
1545 this stage of development, it is only useful for testing kdbus
1546 and should not be used in production. Note: if "--enable-kdbus"
1547 is specified, and the kdbus.ko kernel module is available, and
1548 "kdbus" is added to the kernel command line, the entire system
1549 runs with kdbus instead of dbus-daemon, with the above mentioned
1550 problem of missing the system policy enforcement. Also a future
1551 version of kdbus.ko or a newer systemd will not be compatible with
1552 each other, and will unlikely be able to boot the machine if only
1553 one of them is updated.
1555 * systemctl gained a new "import-environment" command which
1556 uploads the caller's environment (or parts thereof) into the
1557 service manager so that it is inherited by services started
1558 by the manager. This is useful to upload variables like
1559 $DISPLAY into the user service manager.
1561 * A new PrivateDevices= switch has been added to service units
1562 which allows running a service with a namespaced /dev
1563 directory that does not contain any device nodes for
1564 physical devices. More specifically, it only includes devices
1565 such as /dev/null, /dev/urandom, and /dev/zero which are API
1568 * logind has been extended to support behaviour like VT
1569 switching on seats that do not support a VT. This makes
1570 multi-session available on seats that are not the first seat
1571 (seat0), and on systems where kernel support for VTs has
1572 been disabled at compile-time.
1574 * If a process holds a delay lock for system sleep or shutdown
1575 and fails to release it in time, we will now log its
1576 identity. This makes it easier to identify processes that
1577 cause slow suspends or power-offs.
1579 * When parsing /etc/crypttab, support for a new key-slot=
1580 option as supported by Debian is added. It allows indicating
1581 which LUKS slot to use on disk, speeding up key loading.
1583 * The sd_journald_sendv() API call has been checked and
1584 officially declared to be async-signal-safe so that it may
1585 be invoked from signal handlers for logging purposes.
1587 * Boot-time status output is now enabled automatically after a
1588 short timeout if boot does not progress, in order to give
1589 the user an indication what she or he is waiting for.
1591 * The boot-time output has been improved to show how much time
1592 remains until jobs expire.
1594 * The KillMode= switch in service units gained a new possible
1595 value "mixed". If set, and the unit is shut down, then the
1596 initial SIGTERM signal is sent only to the main daemon
1597 process, while the following SIGKILL signal is sent to
1598 all remaining processes of the service.
1600 * When a scope unit is registered, a new property "Controller"
1601 may be set. If set to a valid bus name, systemd will send a
1602 RequestStop() signal to this name when it would like to shut
1603 down the scope. This may be used to hook manager logic into
1604 the shutdown logic of scope units. Also, scope units may now
1605 be put in a special "abandoned" state, in which case the
1606 manager process which created them takes no further
1607 responsibilities for it.
1609 * When reading unit files, systemd will now verify
1610 the access mode of these files, and warn about certain
1611 suspicious combinations. This has been added to make it
1612 easier to track down packaging bugs where unit files are
1613 marked executable or world-writable.
1615 * systemd-nspawn gained a new "--setenv=" switch to set
1616 container-wide environment variables. The similar option in
1617 systemd-activate was renamed from "--environment=" to
1618 "--setenv=" for consistency.
1620 * systemd-nspawn has been updated to create a new kdbus domain
1621 for each container that is invoked, thus allowing each
1622 container to have its own set of system and user buses,
1623 independent of the host.
1625 * systemd-nspawn gained a new --drop-capability= switch to run
1626 the container with less capabilities than the default. Both
1627 --drop-capability= and --capability= now take the special
1628 string "all" for dropping or keeping all capabilities.
1630 * systemd-nspawn gained new switches for executing containers
1631 with specific SELinux labels set.
1633 * systemd-nspawn gained a new --quiet switch to not generate
1634 any additional output but the container's own console
1637 * systemd-nspawn gained a new --share-system switch to run a
1638 container without PID namespacing enabled.
1640 * systemd-nspawn gained a new --register= switch to control
1641 whether the container is registered with systemd-machined or
1642 not. This is useful for containers that do not run full
1643 OS images, but only specific apps.
1645 * systemd-nspawn gained a new --keep-unit which may be used
1646 when invoked as the only program from a service unit, and
1647 results in registration of the unit service itself in
1648 systemd-machined, instead of a newly opened scope unit.
1650 * systemd-nspawn gained a new --network-interface= switch for
1651 moving arbitrary interfaces to the container. The new
1652 --network-veth switch creates a virtual Ethernet connection
1653 between host and container. The new --network-bridge=
1654 switch then allows assigning the host side of this virtual
1655 Ethernet connection to a bridge device.
1657 * systemd-nspawn gained a new --personality= switch for
1658 setting the kernel personality for the container. This is
1659 useful when running a 32-bit container on a 64-bit host. A
1660 similar option Personality= is now also available for service
1663 * logind will now also track a "Desktop" identifier for each
1664 session which encodes the desktop environment of it. This is
1665 useful for desktop environments that want to identify
1666 multiple running sessions of itself easily.
1668 * A new SELinuxContext= setting for service units has been
1669 added that allows setting a specific SELinux execution
1670 context for a service.
1672 * Most systemd client tools will now honour $SYSTEMD_LESS for
1673 settings of the "less" pager. By default, these tools will
1674 override $LESS to allow certain operations to work, such as
1675 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1676 influence this logic.
1678 * systemd's "seccomp" hook-up has been changed to make use of
1679 the libseccomp library instead of using its own
1680 implementation. This has benefits for portability among
1683 * For usage together with SystemCallFilter=, a new
1684 SystemCallErrorNumber= setting has been introduced that
1685 allows configuration of a system error number to be returned
1686 on filtered system calls, instead of immediately killing the
1687 process. Also, SystemCallArchitectures= has been added to
1688 limit access to system calls of a particular architecture
1689 (in order to turn off support for unused secondary
1690 architectures). There is also a global
1691 SystemCallArchitectures= setting in system.conf now to turn
1692 off support for non-native system calls system-wide.
1694 * systemd requires a kernel with a working name_to_handle_at(),
1695 please see the kernel config requirements in the README file.
1697 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1698 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1699 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1700 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1701 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1702 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1703 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1704 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1705 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1706 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1707 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1708 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1709 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1710 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1711 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1712 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1713 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1714 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1715 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1716 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1717 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1718 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1719 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1720 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1722 -- Berlin, 2014-02-20
1726 * logind has gained support for facilitating privileged input
1727 and drm device access for unprivileged clients. This work is
1728 useful to allow Wayland display servers (and similar
1729 programs, such as kmscon) to run under the user's ID and
1730 access input and drm devices which are normally
1731 protected. When this is used (and the kernel is new enough)
1732 logind will "mute" IO on the file descriptors passed to
1733 Wayland as long as it is in the background and "unmute" it
1734 if it returns into the foreground. This allows secure
1735 session switching without allowing background sessions to
1736 eavesdrop on input and display data. This also introduces
1737 session switching support if VT support is turned off in the
1738 kernel, and on seats that are not seat0.
1740 * A new kernel command line option luks.options= is understood
1741 now which allows specifying LUKS options for usage for LUKS
1742 encrypted partitions specified with luks.uuid=.
1744 * tmpfiles.d(5) snippets may now use specifier expansion in
1745 path names. More specifically %m, %b, %H, %v, are now
1746 replaced by the local machine id, boot id, hostname, and
1747 kernel version number.
1749 * A new tmpfiles.d(5) command "m" has been introduced which
1750 may be used to change the owner/group/access mode of a file
1751 or directory if it exists, but do nothing if it does not.
1753 * This release removes high-level support for the
1754 MemorySoftLimit= cgroup setting. The underlying kernel
1755 cgroup attribute memory.soft_limit= is currently badly
1756 designed and likely to be removed from the kernel API in its
1757 current form, hence we should not expose it for now.
1759 * The memory.use_hierarchy cgroup attribute is now enabled for
1760 all cgroups systemd creates in the memory cgroup
1761 hierarchy. This option is likely to be come the built-in
1762 default in the kernel anyway, and the non-hierarchial mode
1763 never made much sense in the intrinsically hierarchial
1766 * A new field _SYSTEMD_SLICE= is logged along with all journal
1767 messages containing the slice a message was generated
1768 from. This is useful to allow easy per-customer filtering of
1769 logs among other things.
1771 * systemd-journald will no longer adjust the group of journal
1772 files it creates to the "systemd-journal" group. Instead we
1773 rely on the journal directory to be owned by the
1774 "systemd-journal" group, and its setgid bit set, so that the
1775 kernel file system layer will automatically enforce that
1776 journal files inherit this group assignment. The reason for
1777 this change is that we cannot allow NSS look-ups from
1778 journald which would be necessary to resolve
1779 "systemd-journal" to a numeric GID, because this might
1780 create deadlocks if NSS involves synchronous queries to
1781 other daemons (such as nscd, or sssd) which in turn are
1782 logging clients of journald and might block on it, which
1783 would then dead lock. A tmpfiles.d(5) snippet included in
1784 systemd will make sure the setgid bit and group are
1785 properly set on the journal directory if it exists on every
1786 boot. However, we recommend adjusting it manually after
1787 upgrades too (or from RPM scriptlets), so that the change is
1788 not delayed until next reboot.
1790 * Backlight and random seed files in /var/lib/ have moved into
1791 the /var/lib/systemd/ directory, in order to centralize all
1792 systemd generated files in one directory.
1794 * Boot time performance measurements (as displayed by
1795 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1796 performance information if that's available to determine how
1797 much time BIOS and boot loader initialization required. With
1798 a sufficiently new BIOS you hence no longer need to boot
1799 with Gummiboot to get access to such information.
1801 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1802 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1803 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1804 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1805 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1806 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1807 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1809 -- Berlin, 2013-10-02
1813 * The Restart= option for services now understands a new
1814 on-watchdog setting, which will restart the service
1815 automatically if the service stops sending out watchdog keep
1816 alive messages (as configured with WatchdogSec=).
1818 * The getty generator (which is responsible for bringing up a
1819 getty on configured serial consoles) will no longer only
1820 start a getty on the primary kernel console but on all
1821 others, too. This makes the order in which console= is
1822 specified on the kernel command line less important.
1824 * libsystemd-logind gained a new sd_session_get_vt() call to
1825 retrieve the VT number of a session.
1827 * If the option "tries=0" is set for an entry of /etc/crypttab
1828 its passphrase is queried indefinitely instead of any
1829 maximum number of tries.
1831 * If a service with a configure PID file terminates its PID
1832 file will now be removed automatically if it still exists
1833 afterwards. This should put an end to stale PID files.
1835 * systemd-run will now also take relative binary path names
1836 for execution and no longer insists on absolute paths.
1838 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1839 paths that are optionally prefixed with "-" to indicate that
1840 it should not be considered a failure if they do not exist.
1842 * journalctl -o (and similar commands) now understands a new
1843 output mode "short-precise", it is similar to "short" but
1844 shows timestamps with usec accuracy.
1846 * The option "discard" (as known from Debian) is now
1847 synonymous to "allow-discards" in /etc/crypttab. In fact,
1848 "discard" is preferred now (since it is easier to remember
1851 * Some licensing clean-ups were made, so that more code is now
1852 LGPL-2.1 licensed than before.
1854 * A minimal tool to save/restore the display backlight
1855 brightness across reboots has been added. It will store the
1856 backlight setting as late as possible at shutdown, and
1857 restore it as early as possible during reboot.
1859 * A logic to automatically discover and enable home and swap
1860 partitions on GPT disks has been added. With this in place
1861 /etc/fstab becomes optional for many setups as systemd can
1862 discover certain partitions located on the root disk
1863 automatically. Home partitions are recognized under their
1864 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1865 partitions are recognized under their GPT type ID
1866 0657fd6da4ab43c484e50933c84b4f4f.
1868 * systemd will no longer pass any environment from the kernel
1869 or initrd to system services. If you want to set an
1870 environment for all services, do so via the kernel command
1871 line systemd.setenv= assignment.
1873 * The systemd-sysctl tool no longer natively reads the file
1874 /etc/sysctl.conf. If desired, the file should be symlinked
1875 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1876 legacy support by a symlink rather than built-in code, it
1877 also makes the otherwise hidden order of application of the
1878 different files visible. (Note that this partly reverts to a
1879 pre-198 application order of sysctl knobs!)
1881 * The "systemctl set-log-level" and "systemctl dump" commands
1882 have been moved to systemd-analyze.
1884 * systemd-run learned the new --remain-after-exit switch,
1885 which causes the scope unit not to be cleaned up
1886 automatically after the process terminated.
1888 * tmpfiles learned a new --exclude-prefix= switch to exclude
1889 certain paths from operation.
1891 * journald will now automatically flush all messages to disk
1892 as soon as a message of the log priorities CRIT, ALERT or
1895 Contributions from: Andrew Cook, Brandon Philips, Christian
1896 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1897 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1898 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1899 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1900 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1901 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1902 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1903 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1904 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1905 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1906 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1907 William Giokas, Zbigniew Jędrzejewski-Szmek
1909 -- Berlin, 2013-09-13
1913 * The documentation has been updated to cover the various new
1914 concepts introduced with 205.
1916 * Unit files now understand the new %v specifier which
1917 resolves to the kernel version string as returned by "uname
1920 * systemctl now supports filtering the unit list output by
1921 load state, active state and sub state, using the new
1924 * "systemctl status" will now show the results of the
1925 condition checks (like ConditionPathExists= and similar) of
1926 the last start attempts of the unit. They are also logged to
1929 * "journalctl -b" may now be used to look for boot output of a
1930 specific boot. Try "journalctl -b -1" for the previous boot,
1931 but the syntax is substantially more powerful.
1933 * "journalctl --show-cursor" has been added which prints the
1934 cursor string the last shown log line. This may then be used
1935 with the new "journalctl --after-cursor=" switch to continue
1936 browsing logs from that point on.
1938 * "journalctl --force" may now be used to force regeneration
1941 * Creation of "dead" device nodes has been moved from udev
1942 into kmod and tmpfiles. Previously, udev would read the kmod
1943 databases to pre-generate dead device nodes based on meta
1944 information contained in kernel modules, so that these would
1945 be auto-loaded on access rather then at boot. As this
1946 does not really have much to do with the exposing actual
1947 kernel devices to userspace this has always been slightly
1948 alien in the udev codebase. Following the new scheme kmod
1949 will now generate a runtime snippet for tmpfiles from the
1950 module meta information and it now is tmpfiles' job to the
1951 create the nodes. This also allows overriding access and
1952 other parameters for the nodes using the usual tmpfiles
1953 facilities. As side effect this allows us to remove the
1954 CAP_SYS_MKNOD capability bit from udevd entirely.
1956 * logind's device ACLs may now be applied to these "dead"
1957 devices nodes too, thus finally allowing managed access to
1958 devices such as /dev/snd/sequencer whithout loading the
1959 backing module right-away.
1961 * A new RPM macro has been added that may be used to apply
1962 tmpfiles configuration during package installation.
1964 * systemd-detect-virt and ConditionVirtualization= now can
1965 detect User-Mode-Linux machines (UML).
1967 * journald will now implicitly log the effective capabilities
1968 set of processes in the message metadata.
1970 * systemd-cryptsetup has gained support for TrueCrypt volumes.
1972 * The initrd interface has been simplified (more specifically,
1973 support for passing performance data via environment
1974 variables and fsck results via files in /run has been
1975 removed). These features were non-essential, and are
1976 nowadays available in a much nicer way by having systemd in
1977 the initrd serialize its state and have the hosts systemd
1978 deserialize it again.
1980 * The udev "keymap" data files and tools to apply keyboard
1981 specific mappings of scan to key codes, and force-release
1982 scan code lists have been entirely replaced by a udev
1983 "keyboard" builtin and a hwdb data file.
1985 * systemd will now honour the kernel's "quiet" command line
1986 argument also during late shutdown, resulting in a
1987 completely silent shutdown when used.
1989 * There's now an option to control the SO_REUSEPORT socket
1990 option in .socket units.
1992 * Instance units will now automatically get a per-template
1993 subslice of system.slice unless something else is explicitly
1994 configured. For example, instances of sshd@.service will now
1995 implicitly be placed in system-sshd.slice rather than
1996 system.slice as before.
1998 * Test coverage support may now be enabled at build time.
2000 Contributions from: Dave Reisner, Frederic Crozat, Harald
2001 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
2002 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
2003 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
2004 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
2005 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
2006 Giokas, Zbigniew Jędrzejewski-Szmek
2008 -- Berlin, 2013-07-23
2012 * Two new unit types have been introduced:
2014 Scope units are very similar to service units, however, are
2015 created out of pre-existing processes -- instead of PID 1
2016 forking off the processes. By using scope units it is
2017 possible for system services and applications to group their
2018 own child processes (worker processes) in a powerful way
2019 which then maybe used to organize them, or kill them
2020 together, or apply resource limits on them.
2022 Slice units may be used to partition system resources in an
2023 hierarchial fashion and then assign other units to them. By
2024 default there are now three slices: system.slice (for all
2025 system services), user.slice (for all user sessions),
2026 machine.slice (for VMs and containers).
2028 Slices and scopes have been introduced primarily in
2029 context of the work to move cgroup handling to a
2030 single-writer scheme, where only PID 1
2031 creates/removes/manages cgroups.
2033 * There's a new concept of "transient" units. In contrast to
2034 normal units these units are created via an API at runtime,
2035 not from configuration from disk. More specifically this
2036 means it is now possible to run arbitrary programs as
2037 independent services, with all execution parameters passed
2038 in via bus APIs rather than read from disk. Transient units
2039 make systemd substantially more dynamic then it ever was,
2040 and useful as a general batch manager.
2042 * logind has been updated to make use of scope and slice units
2043 for managing user sessions. As a user logs in he will get
2044 his own private slice unit, to which all sessions are added
2045 as scope units. We also added support for automatically
2046 adding an instance of user@.service for the user into the
2047 slice. Effectively logind will no longer create cgroup
2048 hierarchies on its own now, it will defer entirely to PID 1
2049 for this by means of scope, service and slice units. Since
2050 user sessions this way become entities managed by PID 1
2051 the output of "systemctl" is now a lot more comprehensive.
2053 * A new mini-daemon "systemd-machined" has been added which
2054 may be used by virtualization managers to register local
2055 VMs/containers. nspawn has been updated accordingly, and
2056 libvirt will be updated shortly. machined will collect a bit
2057 of meta information about the VMs/containers, and assign
2058 them their own scope unit (see above). The collected
2059 meta-data is then made available via the "machinectl" tool,
2060 and exposed in "ps" and similar tools. machined/machinectl
2061 is compile-time optional.
2063 * As discussed earlier, the low-level cgroup configuration
2064 options ControlGroup=, ControlGroupModify=,
2065 ControlGroupPersistent=, ControlGroupAttribute= have been
2066 removed. Please use high-level attribute settings instead as
2067 well as slice units.
2069 * A new bus call SetUnitProperties() has been added to alter
2070 various runtime parameters of a unit. This is primarily
2071 useful to alter cgroup parameters dynamically in a nice way,
2072 but will be extended later on to make more properties
2073 modifiable at runtime. systemctl gained a new set-properties
2074 command that wraps this call.
2076 * A new tool "systemd-run" has been added which can be used to
2077 run arbitrary command lines as transient services or scopes,
2078 while configuring a number of settings via the command
2079 line. This tool is currently very basic, however already
2080 very useful. We plan to extend this tool to even allow
2081 queuing of execution jobs with time triggers from the
2082 command line, similar in fashion to "at".
2084 * nspawn will now inform the user explicitly that kernels with
2085 audit enabled break containers, and suggest the user to turn
2088 * Support for detecting the IMA and AppArmor security
2089 frameworks with ConditionSecurity= has been added.
2091 * journalctl gained a new "-k" switch for showing only kernel
2092 messages, mimicking dmesg output; in addition to "--user"
2093 and "--system" switches for showing only user's own logs
2096 * systemd-delta can now show information about drop-in
2097 snippets extending unit files.
2099 * libsystemd-bus has been substantially updated but is still
2100 not available as public API.
2102 * systemd will now look for the "debug" argument on the kernel
2103 command line and enable debug logging, similar to what
2104 "systemd.log_level=debug" already did before.
2106 * "systemctl set-default", "systemctl get-default" has been
2107 added to configure the default.target symlink, which
2108 controls what to boot into by default.
2110 * "systemctl set-log-level" has been added as a convenient
2111 way to raise and lower systemd logging threshold.
2113 * "systemd-analyze plot" will now show the time the various
2114 generators needed for execution, as well as information
2115 about the unit file loading.
2117 * libsystemd-journal gained a new sd_journal_open_files() call
2118 for opening specific journal files. journactl also gained a
2119 new switch to expose this new functionality. Previously we
2120 only supported opening all files from a directory, or all
2121 files from the system, as opening individual files only is
2122 racy due to journal file rotation.
2124 * systemd gained the new DefaultEnvironment= setting in
2125 /etc/systemd/system.conf to set environment variables for
2128 * If a privileged process logs a journal message with the
2129 OBJECT_PID= field set, then journald will automatically
2130 augment this with additional OBJECT_UID=, OBJECT_GID=,
2131 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2132 system services want to log events about specific client
2133 processes. journactl/systemctl has been updated to make use
2134 of this information if all log messages regarding a specific
2137 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2138 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2139 Reisner, David Coppa, David King, David Strauss, Eelco
2140 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2141 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2142 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2143 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2144 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2145 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2146 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2147 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2148 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2149 Łukasz Stelmach, 장동준
2153 * The Python bindings gained some minimal support for the APIs
2154 exposed by libsystemd-logind.
2156 * ConditionSecurity= gained support for detecting SMACK. Since
2157 this condition already supports SELinux and AppArmor we only
2158 miss IMA for this. Patches welcome!
2160 Contributions from: Karol Lewandowski, Lennart Poettering,
2161 Zbigniew Jędrzejewski-Szmek
2165 * systemd-nspawn will now create /etc/resolv.conf if
2166 necessary, before bind-mounting the host's file onto it.
2168 * systemd-nspawn will now store meta information about a
2169 container on the container's cgroup as extended attribute
2170 fields, including the root directory.
2172 * The cgroup hierarchy has been reworked in many ways. All
2173 objects any of the components systemd creates in the cgroup
2174 tree are now suffixed. More specifically, user sessions are
2175 now placed in cgroups suffixed with ".session", users in
2176 cgroups suffixed with ".user", and nspawn containers in
2177 cgroups suffixed with ".nspawn". Furthermore, all cgroup
2178 names are now escaped in a simple scheme to avoid collision
2179 of userspace object names with kernel filenames. This work
2180 is preparation for making these objects relocatable in the
2181 cgroup tree, in order to allow easy resource partitioning of
2182 these objects without causing naming conflicts.
2184 * systemctl list-dependencies gained the new switches
2185 --plain, --reverse, --after and --before.
2187 * systemd-inhibit now shows the process name of processes that
2188 have taken an inhibitor lock.
2190 * nss-myhostname will now also resolve "localhost"
2191 implicitly. This makes /etc/hosts an optional file and
2192 nicely handles that on IPv6 ::1 maps to both "localhost" and
2195 * libsystemd-logind.so gained a new call
2196 sd_get_machine_names() to enumerate running containers and
2197 VMs (currently only supported by very new libvirt and
2198 nspawn). sd_login_monitor can now be used to watch
2199 VMs/containers coming and going.
2201 * .include is not allowed recursively anymore, and only in
2202 unit files. Usually it is better to use drop-in snippets in
2203 .d/*.conf anyway, as introduced with systemd 198.
2205 * systemd-analyze gained a new "critical-chain" command that
2206 determines the slowest chain of units run during system
2207 boot-up. It is very useful for tracking down where
2208 optimizing boot time is the most beneficial.
2210 * systemd will no longer allow manipulating service paths in
2211 the name=systemd:/system cgroup tree using ControlGroup= in
2212 units. (But is still fine with it in all other dirs.)
2214 * There's a new systemd-nspawn@.service service file that may
2215 be used to easily run nspawn containers as system
2216 services. With the container's root directory in
2217 /var/lib/container/foobar it is now sufficient to run
2218 "systemctl start systemd-nspawn@foobar.service" to boot it.
2220 * systemd-cgls gained a new parameter "--machine" to list only
2221 the processes within a certain container.
2223 * ConditionSecurity= now can check for "apparmor". We still
2224 are lacking checks for SMACK and IMA for this condition
2225 check though. Patches welcome!
2227 * A new configuration file /etc/systemd/sleep.conf has been
2228 added that may be used to configure which kernel operation
2229 systemd is supposed to execute when "suspend", "hibernate"
2230 or "hybrid-sleep" is requested. This makes the new kernel
2231 "freeze" state accessible to the user.
2233 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
2234 the passed argument if applicable.
2236 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
2237 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
2238 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
2239 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
2240 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
2241 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
2242 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
2247 * The output of 'systemctl list-jobs' got some polishing. The
2248 '--type=' argument may now be passed more than once. A new
2249 command 'systemctl list-sockets' has been added which shows
2250 a list of kernel sockets systemd is listening on with the
2251 socket units they belong to, plus the units these socket
2254 * The experimental libsystemd-bus library got substantial
2255 updates to work in conjunction with the (also experimental)
2256 kdbus kernel project. It works well enough to exchange
2257 messages with some sophistication. Note that kdbus is not
2258 ready yet, and the library is mostly an elaborate test case
2259 for now, and not installable.
2261 * systemd gained a new unit 'systemd-static-nodes.service'
2262 that generates static device nodes earlier during boot, and
2263 can run in conjunction with udev.
2265 * libsystemd-login gained a new call sd_pid_get_user_unit()
2266 to retrieve the user systemd unit a process is running
2267 in. This is useful for systems where systemd is used as
2270 * systemd-nspawn now places all containers in the new /machine
2271 top-level cgroup directory in the name=systemd
2272 hierarchy. libvirt will soon do the same, so that we get a
2273 uniform separation of /system, /user and /machine for system
2274 services, user processes and containers/virtual
2275 machines. This new cgroup hierarchy is also useful to stick
2276 stable names to specific container instances, which can be
2277 recognized later this way (this name may be controlled
2278 via systemd-nspawn's new -M switch). libsystemd-login also
2279 gained a new call sd_pid_get_machine_name() to retrieve the
2280 name of the container/VM a specific process belongs to.
2282 * bootchart can now store its data in the journal.
2284 * libsystemd-journal gained a new call
2285 sd_journal_add_conjunction() for AND expressions to the
2286 matching logic. This can be used to express more complex
2287 logical expressions.
2289 * journactl can now take multiple --unit= and --user-unit=
2292 * The cryptsetup logic now understands the "luks.key=" kernel
2293 command line switch for specifying a file to read the
2294 decryption key from. Also, if a configured key file is not
2295 found the tool will now automatically fall back to prompting
2298 * Python systemd.journal module was updated to wrap recently
2299 added functions from libsystemd-journal. The interface was
2300 changed to bring the low level interface in s.j._Reader
2301 closer to the C API, and the high level interface in
2302 s.j.Reader was updated to wrap and convert all data about
2305 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
2306 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
2307 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
2308 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
2309 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
2310 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2314 * journalctl --update-catalog now understands a new --root=
2315 option to operate on catalogs found in a different root
2318 * During shutdown after systemd has terminated all running
2319 services a final killing loop kills all remaining left-over
2320 processes. We will now print the name of these processes
2321 when we send SIGKILL to them, since this usually indicates a
2324 * If /etc/crypttab refers to password files stored on
2325 configured mount points automatic dependencies will now be
2326 generated to ensure the specific mount is established first
2327 before the key file is attempted to be read.
2329 * 'systemctl status' will now show information about the
2330 network sockets a socket unit is listening on.
2332 * 'systemctl status' will also shown information about any
2333 drop-in configuration file for units. (Drop-In configuration
2334 files in this context are files such as
2335 /etc/systemd/systemd/foobar.service.d/*.conf)
2337 * systemd-cgtop now optionally shows summed up CPU times of
2338 cgroups. Press '%' while running cgtop to switch between
2339 percentage and absolute mode. This is useful to determine
2340 which cgroups use up the most CPU time over the entire
2341 runtime of the system. systemd-cgtop has also been updated
2342 to be 'pipeable' for processing with further shell tools.
2344 * 'hostnamectl set-hostname' will now allow setting of FQDN
2347 * The formatting and parsing of time span values has been
2348 changed. The parser now understands fractional expressions
2349 such as "5.5h". The formatter will now output fractional
2350 expressions for all time spans under 1min, i.e. "5.123456s"
2351 rather than "5s 123ms 456us". For time spans under 1s
2352 millisecond values are shown, for those under 1ms
2353 microsecond values are shown. This should greatly improve
2354 all time-related output of systemd.
2356 * libsystemd-login and libsystemd-journal gained new
2357 functions for querying the poll() events mask and poll()
2358 timeout value for integration into arbitrary event
2361 * localectl gained the ability to list available X11 keymaps
2362 (models, layouts, variants, options).
2364 * 'systemd-analyze dot' gained the ability to filter for
2365 specific units via shell-style globs, to create smaller,
2366 more useful graphs. I.e. it is now possible to create simple
2367 graphs of all the dependencies between only target units, or
2368 of all units that Avahi has dependencies with.
2370 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2371 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2372 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2373 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2374 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2375 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2376 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2380 * The boot-time readahead implementation for rotating media
2381 will now read the read-ahead data in multiple passes which
2382 consist of all read requests made in equidistant time
2383 intervals. This means instead of strictly reading read-ahead
2384 data in its physical order on disk we now try to find a
2385 middle ground between physical and access time order.
2387 * /etc/os-release files gained a new BUILD_ID= field for usage
2388 on operating systems that provide continuous builds of OS
2391 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
2392 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
2393 William Douglas, Zbigniew Jędrzejewski-Szmek
2397 * systemd-python gained an API exposing libsystemd-daemon.
2399 * The SMACK setup logic gained support for uploading CIPSO
2402 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
2403 ReadOnlyDirectories= and InaccessibleDirectories= has
2404 changed. The private /tmp and /var/tmp directories are now
2405 shared by all processes of a service (which means
2406 ExecStartPre= may now leave data in /tmp that ExecStart= of
2407 the same service can still access). When a service is
2408 stopped its temporary directories are immediately deleted
2409 (normal clean-up with tmpfiles is still done in addition to
2412 * By default, systemd will now set a couple of sysctl
2413 variables in the kernel: the safe sysrq options are turned
2414 on, IP route verification is turned on, and source routing
2415 disabled. The recently added hardlink and softlink
2416 protection of the kernel is turned on. These settings should
2417 be reasonably safe, and good defaults for all new systems.
2419 * The predictable network naming logic may now be turned off
2420 with a new kernel command line switch: net.ifnames=0.
2422 * A new libsystemd-bus module has been added that implements a
2423 pretty complete D-Bus client library. For details see:
2425 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
2427 * journald will now explicitly flush the journal files to disk
2428 at the latest 5min after each write. The file will then also
2429 be marked offline until the next write. This should increase
2430 reliability in case of a crash. The synchronization delay
2431 can be configured via SyncIntervalSec= in journald.conf.
2433 * There's a new remote-fs-setup.target unit that can be used
2434 to pull in specific services when at least one remote file
2435 system is to be mounted.
2437 * There are new targets timers.target and paths.target as
2438 canonical targets to pull user timer and path units in
2439 from. This complements sockets.target with a similar
2440 purpose for socket units.
2442 * libudev gained a new call udev_device_set_attribute_value()
2443 to set sysfs attributes of a device.
2445 * The udev daemon now sets the default number of worker
2446 processes executed in parallel based on the number of available
2447 CPUs instead of the amount of available RAM. This is supposed
2448 to provide a more reliable default and limit a too aggressive
2449 paralellism for setups with 1000s of devices connected.
2451 Contributions from: Auke Kok, Colin Walters, Cristian
2452 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
2453 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
2454 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
2455 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
2456 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
2457 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
2458 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
2459 Zbigniew Jędrzejewski-Szmek
2463 * Configuration of unit files may now be extended via drop-in
2464 files without having to edit/override the unit files
2465 themselves. More specifically, if the administrator wants to
2466 change one value for a service file foobar.service he can
2467 now do so by dropping in a configuration snippet into
2468 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
2469 will load all these snippets and apply them on top of the
2470 main unit configuration file, possibly extending or
2471 overriding its settings. Using these drop-in snippets is
2472 generally nicer than the two earlier options for changing
2473 unit files locally: copying the files from
2474 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
2475 them there; or creating a new file in /etc/systemd/system/
2476 that incorporates the original one via ".include". Drop-in
2477 snippets into these .d/ directories can be placed in any
2478 directory systemd looks for units in, and the usual
2479 overriding semantics between /usr/lib, /etc and /run apply
2482 * Most unit file settings which take lists of items can now be
2483 reset by assigning the empty string to them. For example,
2484 normally, settings such as Environment=FOO=BAR append a new
2485 environment variable assignment to the environment block,
2486 each time they are used. By assigning Environment= the empty
2487 string the environment block can be reset to empty. This is
2488 particularly useful with the .d/*.conf drop-in snippets
2489 mentioned above, since this adds the ability to reset list
2490 settings from vendor unit files via these drop-ins.
2492 * systemctl gained a new "list-dependencies" command for
2493 listing the dependencies of a unit recursively.
2495 * Inhibitors are now honored and listed by "systemctl
2496 suspend", "systemctl poweroff" (and similar) too, not only
2497 GNOME. These commands will also list active sessions by
2500 * Resource limits (as exposed by the various control group
2501 controllers) can now be controlled dynamically at runtime
2502 for all units. More specifically, you can now use a command
2503 like "systemctl set-cgroup-attr foobar.service cpu.shares
2504 2000" to alter the CPU shares a specific service gets. These
2505 settings are stored persistently on disk, and thus allow the
2506 administrator to easily adjust the resource usage of
2507 services with a few simple commands. This dynamic resource
2508 management logic is also available to other programs via the
2509 bus. Almost any kernel cgroup attribute and controller is
2512 * systemd-vconsole-setup will now copy all font settings to
2513 all allocated VTs, where it previously applied them only to
2516 * libsystemd-login gained the new sd_session_get_tty() API
2519 * This release drops support for a few legacy or
2520 distribution-specific LSB facility names when parsing init
2521 scripts: $x-display-manager, $mail-transfer-agent,
2522 $mail-transport-agent, $mail-transfer-agent, $smtp,
2523 $null. Also, the mail-transfer-agent.target unit backing
2524 this has been removed. Distributions which want to retain
2525 compatibility with this should carry the burden for
2526 supporting this themselves and patch support for these back
2527 in, if they really need to. Also, the facilities $syslog and
2528 $local_fs are now ignored, since systemd does not support
2529 early-boot LSB init scripts anymore, and these facilities
2530 are implied anyway for normal services. syslog.target has
2533 * There are new bus calls on PID1's Manager object for
2534 cancelling jobs, and removing snapshot units. Previously,
2535 both calls were only available on the Job and Snapshot
2538 * systemd-journal-gatewayd gained SSL support.
2540 * The various "environment" files, such as /etc/locale.conf
2541 now support continuation lines with a backslash ("\") as
2542 last character in the line, similarly in style (but different)
2543 to how this is supported in shells.
2545 * For normal user processes the _SYSTEMD_USER_UNIT= field is
2546 now implicitly appended to every log entry logged. systemctl
2547 has been updated to filter by this field when operating on a
2548 user systemd instance.
2550 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
2551 CAP_AUDIT_CONTROL capabilities to the capabilities set for
2552 the container. This makes it easier to boot unmodified
2553 Fedora systems in a container, which however still requires
2554 audit=0 to be passed on the kernel command line. Auditing in
2555 kernel and userspace is unfortunately still too broken in
2556 context of containers, hence we recommend compiling it out
2557 of the kernel or using audit=0. Hopefully this will be fixed
2558 one day for good in the kernel.
2560 * nspawn gained the new --bind= and --bind-ro= parameters to
2561 bind mount specific directories from the host into the
2564 * nspawn will now mount its own devpts file system instance
2565 into the container, in order not to leak pty devices from
2566 the host into the container.
2568 * systemd will now read the firmware boot time performance
2569 information from the EFI variables, if the used boot loader
2570 supports this, and takes it into account for boot performance
2571 analysis via "systemd-analyze". This is currently supported
2572 only in conjunction with Gummiboot, but could be supported
2573 by other boot loaders too. For details see:
2575 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
2577 * A new generator has been added that automatically mounts the
2578 EFI System Partition (ESP) to /boot, if that directory
2579 exists, is empty, and no other file system has been
2580 configured to be mounted there.
2582 * logind will now send out PrepareForSleep(false) out
2583 unconditionally, after coming back from suspend. This may be
2584 used by applications as asynchronous notification for
2585 system resume events.
2587 * "systemctl unlock-sessions" has been added, that allows
2588 unlocking the screens of all user sessions at once, similar
2589 to how "systemctl lock-sessions" already locked all users
2590 sessions. This is backed by a new D-Bus call UnlockSessions().
2592 * "loginctl seat-status" will now show the master device of a
2593 seat. (i.e. the device of a seat that needs to be around for
2594 the seat to be considered available, usually the graphics
2597 * tmpfiles gained a new "X" line type, that allows
2598 configuration of files and directories (with wildcards) that
2599 shall be excluded from automatic cleanup ("aging").
2601 * udev default rules set the device node permissions now only
2602 at "add" events, and do not change them any longer with a
2603 later "change" event.
2605 * The log messages for lid events and power/sleep keypresses
2606 now carry a message ID.
2608 * We now have a substantially larger unit test suite, but this
2609 continues to be work in progress.
2611 * udevadm hwdb gained a new --root= parameter to change the
2612 root directory to operate relative to.
2614 * logind will now issue a background sync() request to the kernel
2615 early at shutdown, so that dirty buffers are flushed to disk early
2616 instead of at the last moment, in order to optimize shutdown
2619 * A new bootctl tool has been added that is an interface for
2620 certain boot loader operations. This is currently a preview
2621 and is likely to be extended into a small mechanism daemon
2622 like timedated, localed, hostnamed, and can be used by
2623 graphical UIs to enumerate available boot options, and
2624 request boot into firmware operations.
2626 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2627 the rest of the package. It also has been updated to work
2628 correctly in initrds.
2630 * Policykit previously has been runtime optional, and is now
2631 also compile time optional via a configure switch.
2633 * systemd-analyze has been reimplemented in C. Also "systemctl
2634 dot" has moved into systemd-analyze.
2636 * "systemctl status" with no further parameters will now print
2637 the status of all active or failed units.
2639 * Operations such as "systemctl start" can now be executed
2640 with a new mode "--irreversible" which may be used to queue
2641 operations that cannot accidentally be reversed by a later
2642 job queuing. This is by default used to make shutdown
2643 requests more robust.
2645 * The Python API of systemd now gained a new module for
2646 reading journal files.
2648 * A new tool kernel-install has been added that can install
2649 kernel images according to the Boot Loader Specification:
2651 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2653 * Boot time console output has been improved to provide
2654 animated boot time output for hanging jobs.
2656 * A new tool systemd-activate has been added which can be used
2657 to test socket activation with, directly from the command
2658 line. This should make it much easier to test and debug
2659 socket activation in daemons.
2661 * journalctl gained a new "--reverse" (or -r) option to show
2662 journal output in reverse order (i.e. newest line first).
2664 * journalctl gained a new "--pager-end" (or -e) option to jump
2665 to immediately jump to the end of the journal in the
2666 pager. This is only supported in conjunction with "less".
2668 * journalctl gained a new "--user-unit=" option, that works
2669 similarly to "--unit=" but filters for user units rather than
2672 * A number of unit files to ease adoption of systemd in
2673 initrds has been added. This moves some minimal logic from
2674 the various initrd implementations into systemd proper.
2676 * The journal files are now owned by a new group
2677 "systemd-journal", which exists specifically to allow access
2678 to the journal, and nothing else. Previously, we used the
2679 "adm" group for that, which however possibly covers more
2680 than just journal/log file access. This new group is now
2681 already used by systemd-journal-gatewayd to ensure this
2682 daemon gets access to the journal files and as little else
2683 as possible. Note that "make install" will also set FS ACLs
2684 up for /var/log/journal to give "adm" and "wheel" read
2685 access to it, in addition to "systemd-journal" which owns
2686 the journal files. We recommend that packaging scripts also
2687 add read access to "adm" + "wheel" to /var/log/journal, and
2688 all existing/future journal files. To normal users and
2689 administrators little changes, however packagers need to
2690 ensure to create the "systemd-journal" system group at
2691 package installation time.
2693 * The systemd-journal-gatewayd now runs as unprivileged user
2694 systemd-journal-gateway:systemd-journal-gateway. Packaging
2695 scripts need to create these system user/group at
2698 * timedated now exposes a new boolean property CanNTP that
2699 indicates whether a local NTP service is available or not.
2701 * systemd-detect-virt will now also detect xen PVs
2703 * The pstore file system is now mounted by default, if it is
2706 * In addition to the SELinux and IMA policies we will now also
2707 load SMACK policies at early boot.
2709 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2710 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2711 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2712 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2713 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2714 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2715 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2716 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2717 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2718 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2719 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2720 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2721 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2722 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2726 * Timer units now support calendar time events in addition to
2727 monotonic time events. That means you can now trigger a unit
2728 based on a calendar time specification such as "Thu,Fri
2729 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2730 or fifth day of any month of the year 2013, given that it is
2731 a thursday or friday. This brings timer event support
2732 considerably closer to cron's capabilities. For details on
2733 the supported calendar time specification language see
2736 * udev now supports a number of different naming policies for
2737 network interfaces for predictable names, and a combination
2738 of these policies is now the default. Please see this wiki
2739 document for details:
2741 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2743 * Auke Kok's bootchart implementation has been added to the
2744 systemd tree. It is an optional component that can graph the
2745 boot in quite some detail. It is one of the best bootchart
2746 implementations around and minimal in its code and
2749 * nss-myhostname has been integrated into the systemd source
2750 tree. nss-myhostname guarantees that the local hostname
2751 always stays resolvable via NSS. It has been a weak
2752 requirement of systemd-hostnamed since a long time, and
2753 since its code is actually trivial we decided to just
2754 include it in systemd's source tree. It can be turned off
2755 with a configure switch.
2757 * The read-ahead logic is now capable of properly detecting
2758 whether a btrfs file system is on SSD or rotating media, in
2759 order to optimize the read-ahead scheme. Previously, it was
2760 only capable of detecting this on traditional file systems
2763 * In udev, additional device properties are now read from the
2764 IAB in addition to the OUI database. Also, Bluetooth company
2765 identities are attached to the devices as well.
2767 * In service files %U may be used as specifier that is
2768 replaced by the configured user name of the service.
2770 * nspawn may now be invoked without a controlling TTY. This
2771 makes it suitable for invocation as its own service. This
2772 may be used to set up a simple containerized server system
2773 using only core OS tools.
2775 * systemd and nspawn can now accept socket file descriptors
2776 when they are started for socket activation. This enables
2777 implementation of socket activated nspawn
2778 containers. i.e. think about autospawning an entire OS image
2779 when the first SSH or HTTP connection is received. We expect
2780 that similar functionality will also be added to libvirt-lxc
2783 * journalctl will now suppress ANSI color codes when
2784 presenting log data.
2786 * systemctl will no longer show control group information for
2787 a unit if a the control group is empty anyway.
2789 * logind can now automatically suspend/hibernate/shutdown the
2792 * /etc/machine-info and hostnamed now also expose the chassis
2793 type of the system. This can be used to determine whether
2794 the local system is a laptop, desktop, handset or
2795 tablet. This information may either be configured by the
2796 user/vendor or is automatically determined from ACPI and DMI
2797 information if possible.
2799 * A number of PolicyKit actions are now bound together with
2800 "imply" rules. This should simplify creating UIs because
2801 many actions will now authenticate similar ones as well.
2803 * Unit files learnt a new condition ConditionACPower= which
2804 may be used to conditionalize a unit depending on whether an
2805 AC power source is connected or not, of whether the system
2806 is running on battery power.
2808 * systemctl gained a new "is-failed" verb that may be used in
2809 shell scripts and suchlike to check whether a specific unit
2810 is in the "failed" state.
2812 * The EnvironmentFile= setting in unit files now supports file
2813 globbing, and can hence be used to easily read a number of
2814 environment files at once.
2816 * systemd will no longer detect and recognize specific
2817 distributions. All distribution-specific #ifdeffery has been
2818 removed, systemd is now fully generic and
2819 distribution-agnostic. Effectively, not too much is lost as
2820 a lot of the code is still accessible via explicit configure
2821 switches. However, support for some distribution specific
2822 legacy configuration file formats has been dropped. We
2823 recommend distributions to simply adopt the configuration
2824 files everybody else uses now and convert the old
2825 configuration from packaging scripts. Most distributions
2826 already did that. If that's not possible or desirable,
2827 distributions are welcome to forward port the specific
2828 pieces of code locally from the git history.
2830 * When logging a message about a unit systemd will now always
2831 log the unit name in the message meta data.
2833 * localectl will now also discover system locale data that is
2834 not stored in locale archives, but directly unpacked.
2836 * logind will no longer unconditionally use framebuffer
2837 devices as seat masters, i.e. as devices that are required
2838 to be existing before a seat is considered preset. Instead,
2839 it will now look for all devices that are tagged as
2840 "seat-master" in udev. By default framebuffer devices will
2841 be marked as such, but depending on local systems other
2842 devices might be marked as well. This may be used to
2843 integrate graphics cards using closed source drivers (such
2844 as NVidia ones) more nicely into logind. Note however, that
2845 we recommend using the open source NVidia drivers instead,
2846 and no udev rules for the closed-source drivers will be
2847 shipped from us upstream.
2849 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2850 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2851 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2852 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2853 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2854 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2855 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2856 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2857 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2858 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2859 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2860 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2865 * udev gained support for loading additional device properties
2866 from an indexed database that is keyed by vendor/product IDs
2867 and similar device identifiers. For the beginning this
2868 "hwdb" is populated with data from the well-known PCI and
2869 USB database, but also includes PNP, ACPI and OID data. In
2870 the longer run this indexed database shall grow into
2871 becoming the one central database for non-essential
2872 userspace device metadata. Previously, data from the PCI/USB
2873 database was only attached to select devices, since the
2874 lookup was a relatively expensive operation due to O(n) time
2875 complexity (with n being the number of entries in the
2876 database). Since this is now O(1), we decided to add in this
2877 data for all devices where this is available, by
2878 default. Note that the indexed database needs to be rebuilt
2879 when new data files are installed. To achieve this you need
2880 to update your packaging scripts to invoke "udevadm hwdb
2881 --update" after installation of hwdb data files. For
2882 RPM-based distributions we introduced the new
2883 %udev_hwdb_update macro for this purpose.
2885 * The Journal gained support for the "Message Catalog", an
2886 indexed database to link up additional information with
2887 journal entries. For further details please check:
2889 http://www.freedesktop.org/wiki/Software/systemd/catalog
2891 The indexed message catalog database also needs to be
2892 rebuilt after installation of message catalog files. Use
2893 "journalctl --update-catalog" for this. For RPM-based
2894 distributions we introduced the %journal_catalog_update
2895 macro for this purpose.
2897 * The Python Journal bindings gained support for the standard
2898 Python logging framework.
2900 * The Journal API gained new functions for checking whether
2901 the underlying file system of a journal file is capable of
2902 properly reporting file change notifications, or whether
2903 applications that want to reflect journal changes "live"
2904 need to recheck journal files continuously in appropriate
2907 * It is now possible to set the "age" field for tmpfiles
2908 entries to 0, indicating that files matching this entry
2909 shall always be removed when the directories are cleaned up.
2911 * coredumpctl gained a new "gdb" verb which invokes gdb
2912 right-away on the selected coredump.
2914 * There's now support for "hybrid sleep" on kernels that
2915 support this, in addition to "suspend" and "hibernate". Use
2916 "systemctl hybrid-sleep" to make use of this.
2918 * logind's HandleSuspendKey= setting (and related settings)
2919 now gained support for a new "lock" setting to simply
2920 request the screen lock on all local sessions, instead of
2921 actually executing a suspend or hibernation.
2923 * systemd will now mount the EFI variables file system by
2926 * Socket units now gained support for configuration of the
2927 SMACK security label.
2929 * timedatectl will now output the time of the last and next
2930 daylight saving change.
2932 * We dropped support for various legacy and distro-specific
2933 concepts, such as insserv, early-boot SysV services
2934 (i.e. those for non-standard runlevels such as 'b' or 'S')
2935 or ArchLinux /etc/rc.conf support. We recommend the
2936 distributions who still need support this to either continue
2937 to maintain the necessary patches downstream, or find a
2938 different solution. (Talk to us if you have questions!)
2940 * Various systemd components will now bypass PolicyKit checks
2941 for root and otherwise handle properly if PolicyKit is not
2942 found to be around. This should fix most issues for
2943 PolicyKit-less systems. Quite frankly this should have been
2944 this way since day one. It is absolutely our intention to
2945 make systemd work fine on PolicyKit-less systems, and we
2946 consider it a bug if something does not work as it should if
2947 PolicyKit is not around.
2949 * For embedded systems it is now possible to build udev and
2950 systemd without blkid and/or kmod support.
2952 * "systemctl switch-root" is now capable of switching root
2953 more than once. I.e. in addition to transitions from the
2954 initrd to the host OS it is now possible to transition to
2955 further OS images from the host. This is useful to implement
2956 offline updating tools.
2958 * Various other additions have been made to the RPM macros
2959 shipped with systemd. Use %udev_rules_update() after
2960 installing new udev rules files. %_udevhwdbdir,
2961 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
2962 %_sysctldir are now available which resolve to the right
2963 directories for packages to place various data files in.
2965 * journalctl gained the new --full switch (in addition to
2966 --all, to disable ellipsation for long messages.
2968 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
2969 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
2970 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
2971 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
2972 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
2973 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
2974 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
2975 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
2976 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
2980 * journalctl gained new --since= and --until= switches to
2981 filter by time. It also now supports nice filtering for
2982 units via --unit=/-u.
2984 * Type=oneshot services may use ExecReload= and do the
2987 * The journal daemon now supports time-based rotation and
2988 vacuuming, in addition to the usual disk-space based
2991 * The journal will now index the available field values for
2992 each field name. This enables clients to show pretty drop
2993 downs of available match values when filtering. The bash
2994 completion of journalctl has been updated
2995 accordingly. journalctl gained a new switch -F to list all
2996 values a certain field takes in the journal database.
2998 * More service events are now written as structured messages
2999 to the journal, and made recognizable via message IDs.
3001 * The timedated, localed and hostnamed mini-services which
3002 previously only provided support for changing time, locale
3003 and hostname settings from graphical DEs such as GNOME now
3004 also have a minimal (but very useful) text-based client
3005 utility each. This is probably the nicest way to changing
3006 these settings from the command line now, especially since
3007 it lists available options and is fully integrated with bash
3010 * There's now a new tool "systemd-coredumpctl" to list and
3011 extract coredumps from the journal.
3013 * We now install a README each in /var/log/ and
3014 /etc/rc.d/init.d explaining where the system logs and init
3015 scripts went. This hopefully should help folks who go to
3016 that dirs and look into the otherwise now empty void and
3017 scratch their heads.
3019 * When user-services are invoked (by systemd --user) the
3020 $MANAGERPID env var is set to the PID of systemd.
3022 * SIGRTMIN+24 when sent to a --user instance will now result
3023 in immediate termination of systemd.
3025 * gatewayd received numerous feature additions such as a
3026 "follow" mode, for live syncing and filtering.
3028 * browse.html now allows filtering and showing detailed
3029 information on specific entries. Keyboard navigation and
3030 mouse screen support has been added.
3032 * gatewayd/journalctl now supports HTML5/JSON
3033 Server-Sent-Events as output.
3035 * The SysV init script compatibility logic will now
3036 heuristically determine whether a script supports the
3037 "reload" verb, and only then make this available as
3040 * "systemctl status --follow" has been removed, use "journalctl
3043 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3044 have been removed since they are hardly useful to be
3047 * And I'd like to take the opportunity to specifically mention
3048 Zbigniew for his great contributions. Zbigniew, you rock!
3050 Contributions from: Andrew Eikum, Christian Hesse, Colin
3051 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3052 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3053 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3054 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3055 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3056 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3060 * If /etc/vconsole.conf is non-existent or empty we will no
3061 longer load any console font or key map at boot by
3062 default. Instead the kernel defaults will be left
3063 intact. This is definitely the right thing to do, as no
3064 configuration should mean no configuration, and hard-coding
3065 font names that are different on all archs is probably a bad
3066 idea. Also, the kernel default key map and font should be
3067 good enough for most cases anyway, and mostly identical to
3068 the userspace fonts/key maps we previously overloaded them
3069 with. If distributions want to continue to default to a
3070 non-kernel font or key map they should ship a default
3071 /etc/vconsole.conf with the appropriate contents.
3073 Contributions from: Colin Walters, Daniel J Walsh, Dave
3074 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3075 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3079 * journalctl gained a new --cursor= switch to show entries
3080 starting from the specified location in the journal.
3082 * We now enforce a size limit on journal entry fields exported
3083 with "-o json" in journalctl. Fields larger than 4K will be
3084 assigned null. This can be turned off with --all.
3086 * An (optional) journal gateway daemon is now available as
3087 "systemd-journal-gatewayd.service". This service provides
3088 access to the journal via HTTP and JSON. This functionality
3089 will be used to implement live log synchronization in both
3090 pull and push modes, but has various other users too, such
3091 as easy log access for debugging of embedded devices. Right
3092 now it is already useful to retrieve the journal via HTTP:
3094 # systemctl start systemd-journal-gatewayd.service
3095 # wget http://localhost:19531/entries
3097 This will download the journal contents in a
3098 /var/log/messages compatible format. The same as JSON:
3100 # curl -H"Accept: application/json" http://localhost:19531/entries
3102 This service is also accessible via a web browser where a
3103 single static HTML5 app is served that uses the JSON logic
3104 to enable the user to do some basic browsing of the
3105 journal. This will be extended later on. Here's an example
3106 screenshot of this app in its current state:
3108 http://0pointer.de/public/journal-gatewayd
3110 Contributions from: Kay Sievers, Lennart Poettering, Robert
3111 Milasan, Tom Gundersen
3115 * The bash completion logic is now available for journalctl
3118 * We do not mount the "cpuset" controller anymore together with
3119 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
3120 started if no parameters are assigned to it. "cpuset" hence
3121 broke code that assumed it it could create "cpu" groups and
3124 * journalctl -f will now subscribe to terminal size changes,
3125 and line break accordingly.
3127 Contributions from: Dave Reisner, Kay Sievers, Lennart
3128 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3132 * nspawn will now create a symlink /etc/localtime in the
3133 container environment, copying the host's timezone
3134 setting. Previously this has been done via a bind mount, but
3135 since symlinks cannot be bind mounted this has now been
3136 changed to create/update the appropriate symlink.
3138 * journalctl -n's line number argument is now optional, and
3139 will default to 10 if omitted.
3141 * journald will now log the maximum size the journal files may
3142 take up on disk. This is particularly useful if the default
3143 built-in logic of determining this parameter from the file
3144 system size is used. Use "systemctl status
3145 systemd-journald.service" to see this information.
3147 * The multi-seat X wrapper tool has been stripped down. As X
3148 is now capable of enumerating graphics devices via udev in a
3149 seat-aware way the wrapper is not strictly necessary
3150 anymore. A stripped down temporary stop-gap is still shipped
3151 until the upstream display managers have been updated to
3152 fully support the new X logic. Expect this wrapper to be
3153 removed entirely in one of the next releases.
3155 * HandleSleepKey= in logind.conf has been split up into
3156 HandleSuspendKey= and HandleHibernateKey=. The old setting
3157 is not available anymore. X11 and the kernel are
3158 distuingishing between these keys and we should too. This
3159 also means the inhibition lock for these keys has been split
3162 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
3163 Poettering, Lukas Nykryn, Václav Pavlín
3167 * Whenever a unit changes state we will now log this to the
3168 journal and show along the unit's own log output in
3171 * ConditionPathIsMountPoint= can now properly detect bind
3172 mount points too. (Previously, a bind mount of one file
3173 system to another place in the same file system could not be
3174 detected as mount, since they shared struct stat's st_dev
3177 * We will now mount the cgroup controllers cpu, cpuacct,
3178 cpuset and the controllers net_cls, net_prio together by
3181 * nspawn containers will now have a virtualized boot
3182 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
3183 over with a randomized ID at container initialization). This
3184 has the effect of making "journalctl -b" do the right thing
3187 * The JSON output journal serialization has been updated not
3188 to generate "endless" list objects anymore, but rather one
3189 JSON object per line. This is more in line how most JSON
3190 parsers expect JSON objects. The new output mode
3191 "json-pretty" has been added to provide similar output, but
3192 neatly aligned for readability by humans.
3194 * We dropped all explicit sync() invocations in the shutdown
3195 code. The kernel does this implicitly anyway in the kernel
3196 reboot() syscall. halt(8)'s -n option is now a compatibility
3199 * We now support virtualized reboot() in containers, as
3200 supported by newer kernels. We will fall back to exit() if
3201 CAP_SYS_REBOOT is not available to the container. Also,
3202 nspawn makes use of this now and will actually reboot the
3203 container if the containerized OS asks for that.
3205 * journalctl will only show local log output by default
3206 now. Use --merge (-m) to show remote log output, too.
3208 * libsystemd-journal gained the new sd_journal_get_usage()
3209 call to determine the current disk usage of all journal
3210 files. This is exposed in the new "journalctl --disk-usage"
3213 * journald gained a new configuration setting SplitMode= in
3214 journald.conf which may be used to control how user journals
3215 are split off. See journald.conf(5) for details.
3217 * A new condition type ConditionFileNotEmpty= has been added.
3219 * tmpfiles' "w" lines now support file globbing, to write
3220 multiple files at once.
3222 * We added Python bindings for the journal submission
3223 APIs. More Python APIs for a number of selected APIs will
3224 likely follow. Note that we intend to add native bindings
3225 only for the Python language, as we consider it common
3226 enough to deserve bindings shipped within systemd. There are
3227 various projects outside of systemd that provide bindings
3228 for languages such as PHP or Lua.
3230 * Many conditions will now resolve specifiers such as %i. In
3231 addition, PathChanged= and related directives of .path units
3232 now support specifiers as well.
3234 * There's now a new RPM macro definition for the system preset
3237 * journald will now warn if it ca not forward a message to the
3238 syslog daemon because its socket is full.
3240 * timedated will no longer write or process /etc/timezone,
3241 except on Debian. As we do not support late mounted /usr
3242 anymore /etc/localtime always being a symlink is now safe,
3243 and hence the information in /etc/timezone is not necessary
3246 * logind will now always reserve one VT for a text getty (VT6
3247 by default). Previously if more than 6 X sessions where
3248 started they took up all the VTs with auto-spawned gettys,
3249 so that no text gettys were available anymore.
3251 * udev will now automatically inform the btrfs kernel logic
3252 about btrfs RAID components showing up. This should make
3253 simple hotplug based btrfs RAID assembly work.
3255 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
3256 (but not for its children which will stay at the kernel
3257 default). This should allow setups with a lot more listening
3260 * systemd will now always pass the configured timezone to the
3261 kernel at boot. timedated will do the same when the timezone
3264 * logind's inhibition logic has been updated. By default,
3265 logind will now handle the lid switch, the power and sleep
3266 keys all the time, even in graphical sessions. If DEs want
3267 to handle these events on their own they should take the new
3268 handle-power-key, handle-sleep-key and handle-lid-switch
3269 inhibitors during their runtime. A simple way to achiveve
3270 that is to invoke the DE wrapped in an invocation of:
3272 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
3274 * Access to unit operations is now checked via SELinux taking
3275 the unit file label and client process label into account.
3277 * systemd will now notify the administrator in the journal
3278 when he over-mounts a non-empty directory.
3280 * There are new specifiers that are resolved in unit files,
3281 for the host name (%H), the machine ID (%m) and the boot ID
3284 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
3285 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
3286 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
3287 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
3288 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
3289 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3290 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
3294 * Support for reading structured kernel messages from
3295 /dev/kmsg has now been added and is enabled by default.
3297 * Support for reading kernel messages from /proc/kmsg has now
3298 been removed. If you want kernel messages in the journal
3299 make sure to run a recent kernel (>= 3.5) that supports
3300 reading structured messages from /dev/kmsg (see
3301 above). /proc/kmsg is now exclusive property of classic
3302 syslog daemons again.
3304 * The libudev API gained the new
3305 udev_device_new_from_device_id() call.
3307 * The logic for file system namespace (ReadOnlyDirectory=,
3308 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
3309 require pivot_root() anymore. This means fewer temporary
3310 directories are created below /tmp for this feature.
3312 * nspawn containers will now see and receive all submounts
3313 made on the host OS below the root file system of the
3316 * Forward Secure Sealing is now supported for Journal files,
3317 which provide cryptographical sealing of journal files so
3318 that attackers cannot alter log history anymore without this
3319 being detectable. Lennart will soon post a blog story about
3320 this explaining it in more detail.
3322 * There are two new service settings RestartPreventExitStatus=
3323 and SuccessExitStatus= which allow configuration of exit
3324 status (exit code or signal) which will be excepted from the
3325 restart logic, resp. consider successful.
3327 * journalctl gained the new --verify switch that can be used
3328 to check the integrity of the structure of journal files and
3329 (if Forward Secure Sealing is enabled) the contents of
3332 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3333 and similar symlinks pre-created. This makes running shells
3334 as container init process a lot more fun.
3336 * The fstab support can now handle PARTUUID= and PARTLABEL=
3339 * A new ConditionHost= condition has been added to match
3340 against the hostname (with globs) and machine ID. This is
3341 useful for clusters where a single OS image is used to
3342 provision a large number of hosts which shall run slightly
3343 different sets of services.
3345 * Services which hit the restart limit will now be placed in a
3348 Contributions from: Bertram Poettering, Dave Reisner, Huang
3349 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3350 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3354 * When running in --user mode systemd will now become a
3355 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3356 tree a lot more organized.
3358 * A new PartOf= unit dependency type has been introduced that
3359 may be used to group services in a natural way.
3361 * "systemctl enable" may now be used to enable instances of
3364 * journalctl now prints error log levels in red, and
3365 warning/notice log levels in bright white. It also supports
3366 filtering by log level now.
3368 * cgtop gained a new -n switch (similar to top), to configure
3369 the maximum number of iterations to run for. It also gained
3370 -b, to run in batch mode (accepting no input).
3372 * The suffix ".service" may now be omitted on most systemctl
3373 command lines involving service unit names.
3375 * There's a new bus call in logind to lock all sessions, as
3376 well as a loginctl verb for it "lock-sessions".
3378 * libsystemd-logind.so gained a new call sd_journal_perror()
3379 that works similar to libc perror() but logs to the journal
3380 and encodes structured information about the error number.
3382 * /etc/crypttab entries now understand the new keyfile-size=
3385 * shutdown(8) now can send a (configurable) wall message when
3386 a shutdown is cancelled.
3388 * The mount propagation mode for the root file system will now
3389 default to "shared", which is useful to make containers work
3390 nicely out-of-the-box so that they receive new mounts from
3391 the host. This can be undone locally by running "mount
3392 --make-rprivate /" if needed.
3394 * The prefdm.service file has been removed. Distributions
3395 should maintain this unit downstream if they intend to keep
3396 it around. However, we recommend writing normal unit files
3397 for display managers instead.
3399 * Since systemd is a crucial part of the OS we will now
3400 default to a number of compiler switches that improve
3401 security (hardening) such as read-only relocations, stack
3402 protection, and suchlike.
3404 * The TimeoutSec= setting for services is now split into
3405 TimeoutStartSec= and TimeoutStopSec= to allow configuration
3406 of individual time outs for the start and the stop phase of
3409 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
3410 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
3411 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
3412 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
3413 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
3414 Gundersen, Zbigniew Jędrzejewski-Szmek
3418 * The journal and id128 C APIs are now fully documented as man
3421 * Extra safety checks have been added when transitioning from
3422 the initial RAM disk to the main system to avoid accidental
3425 * /etc/crypttab entries now understand the new keyfile-offset=
3428 * systemctl -t can now be used to filter by unit load state.
3430 * The journal C API gained the new sd_journal_wait() call to
3431 make writing synchronous journal clients easier.
3433 * journalctl gained the new -D switch to show journals from a
3436 * journalctl now displays a special marker between log
3437 messages of two different boots.
3439 * The journal is now explicitly flushed to /var via a service
3440 systemd-journal-flush.service, rather than implicitly simply
3441 by seeing /var/log/journal to be writable.
3443 * journalctl (and the journal C APIs) can now match for much
3444 more complex expressions, with alternatives and
3447 * When transitioning from the initial RAM disk to the main
3448 system we will now kill all processes in a killing spree to
3449 ensure no processes stay around by accident.
3451 * Three new specifiers may be used in unit files: %u, %h, %s
3452 resolve to the user name, user home directory resp. user
3453 shell. This is useful for running systemd user instances.
3455 * We now automatically rotate journal files if their data
3456 object hash table gets a fill level > 75%. We also size the
3457 hash table based on the configured maximum file size. This
3458 together should lower hash collisions drastically and thus
3459 speed things up a bit.
3461 * journalctl gained the new "--header" switch to introspect
3462 header data of journal files.
3464 * A new setting SystemCallFilters= has been added to services
3465 which may be used to apply blacklists or whitelists to
3466 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
3468 * nspawn gained a new --link-journal= switch (and quicker: -j)
3469 to link the container journal with the host. This makes it
3470 very easy to centralize log viewing on the host for all
3471 guests while still keeping the journal files separated.
3473 * Many bugfixes and optimizations
3475 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
3476 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
3477 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
3482 * Several tools now understand kernel command line arguments,
3483 which are only read when run in an initial RAM disk. They
3484 usually follow closely their normal counterparts, but are
3487 * There's a new tool to analyze the readahead files that are
3488 automatically generated at boot. Use:
3490 /usr/lib/systemd/systemd-readahead analyze /.readahead
3492 * We now provide an early debug shell on tty9 if this enabled. Use:
3494 systemctl enable debug-shell.service
3496 * All plymouth related units have been moved into the Plymouth
3497 package. Please make sure to upgrade your Plymouth version
3500 * systemd-tmpfiles now supports getting passed the basename of
3501 a configuration file only, in which case it will look for it
3502 in all appropriate directories automatically.
3504 * udevadm info now takes a /dev or /sys path as argument, and
3505 does the right thing. Example:
3507 udevadm info /dev/sda
3508 udevadm info /sys/class/block/sda
3510 * systemctl now prints a warning if a unit is stopped but a
3511 unit that might trigger it continues to run. Example: a
3512 service is stopped but the socket that activates it is left
3515 * "systemctl status" will now mention if the log output was
3516 shortened due to rotation since a service has been started.
3518 * The journal API now exposes functions to determine the
3519 "cutoff" times due to rotation.
3521 * journald now understands SIGUSR1 and SIGUSR2 for triggering
3522 immediately flushing of runtime logs to /var if possible,
3523 resp. for triggering immediate rotation of the journal
3526 * It is now considered an error if a service is attempted to
3527 be stopped that is not loaded.
3529 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
3531 * systemd-analyze now supports Python 3
3533 * tmpfiles now supports cleaning up directories via aging
3534 where the first level dirs are always kept around but
3535 directories beneath it automatically aged. This is enabled
3536 by prefixing the age field with '~'.
3538 * Seat objects now expose CanGraphical, CanTTY properties
3539 which is required to deal with very fast bootups where the
3540 display manager might be running before the graphics drivers
3541 completed initialization.
3543 * Seat objects now expose a State property.
3545 * We now include RPM macros for service enabling/disabling
3546 based on the preset logic. We recommend RPM based
3547 distributions to make use of these macros if possible. This
3548 makes it simpler to reuse RPM spec files across
3551 * We now make sure that the collected systemd unit name is
3552 always valid when services log to the journal via
3555 * There's a new man page kernel-command-line(7) detailing all
3556 command line options we understand.
3558 * The fstab generator may now be disabled at boot by passing
3559 fstab=0 on the kernel command line.
3561 * A new kernel command line option modules-load= is now understood
3562 to load a specific kernel module statically, early at boot.
3564 * Unit names specified on the systemctl command line are now
3565 automatically escaped as needed. Also, if file system or
3566 device paths are specified they are automatically turned
3567 into the appropriate mount or device unit names. Example:
3569 systemctl status /home
3570 systemctl status /dev/sda
3572 * The SysVConsole= configuration option has been removed from
3573 system.conf parsing.
3575 * The SysV search path is no longer exported on the D-Bus
3578 * The Names= option is been removed from unit file parsing.
3580 * There's a new man page bootup(7) detailing the boot process.
3582 * Every unit and every generator we ship with systemd now
3583 comes with full documentation. The self-explanatory boot is
3586 * A couple of services gained "systemd-" prefixes in their
3587 name if they wrap systemd code, rather than only external
3588 code. Among them fsck@.service which is now
3589 systemd-fsck@.service.
3591 * The HaveWatchdog property has been removed from the D-Bus
3594 * systemd.confirm_spawn= on the kernel command line should now
3597 * There's a new man page crypttab(5) which details all options
3598 we actually understand.
3600 * systemd-nspawn gained a new --capability= switch to pass
3601 additional capabilities to the container.
3603 * timedated will now read known NTP implementation unit names
3604 from /usr/lib/systemd/ntp-units.d/*.list,
3605 systemd-timedated-ntp.target has been removed.
3607 * journalctl gained a new switch "-b" that lists log data of
3608 the current boot only.
3610 * The notify socket is in the abstract namespace again, in
3611 order to support daemons which chroot() at start-up.
3613 * There is a new Storage= configuration option for journald
3614 which allows configuration of where log data should go. This
3615 also provides a way to disable journal logging entirely, so
3616 that data collected is only forwarded to the console, the
3617 kernel log buffer or another syslog implementation.
3619 * Many bugfixes and optimizations
3621 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3622 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3623 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3624 Shawn Landden, Tom Gundersen
3628 * "systemctl help <unit>" now shows the man page if one is
3631 * Several new man pages have been added.
3633 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3634 MaxLevelConsole= can now be specified in
3635 journald.conf. These options allow reducing the amount of
3636 data stored on disk or forwarded by the log level.
3638 * TimerSlackNSec= can now be specified in system.conf for
3639 PID1. This allows system-wide power savings.
3641 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3642 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3647 * logind is now capable of (optionally) handling power and
3648 sleep keys as well as the lid switch.
3650 * journalctl now understands the syntax "journalctl
3651 /usr/bin/avahi-daemon" to get all log output of a specific
3654 * CapabilityBoundingSet= in system.conf now also influences
3655 the capability bound set of usermode helpers of the kernel.
3657 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3658 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3659 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3660 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3664 * Note that we skipped 139 releases here in order to set the
3665 new version to something that is greater than both udev's
3666 and systemd's most recent version number.
3668 * udev: all udev sources are merged into the systemd source tree now.
3669 All future udev development will happen in the systemd tree. It
3670 is still fully supported to use the udev daemon and tools without
3671 systemd running, like in initramfs or other init systems. Building
3672 udev though, will require the *build* of the systemd tree, but
3673 udev can be properly *run* without systemd.
3675 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3676 should be used to create dead device nodes as workarounds for broken
3679 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3680 no longer supported. udev_monitor_new_from_netlink() needs to be
3681 used to subscribe to events.
3683 * udev: when udevd is started by systemd, processes which are left
3684 behind by forking them off of udev rules, are unconditionally cleaned
3685 up and killed now after the event handling has finished. Services or
3686 daemons must be started as systemd services. Services can be
3687 pulled-in by udev to get started, but they can no longer be directly
3688 forked by udev rules.
3690 * udev: the daemon binary is called systemd-udevd now and installed
3691 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3692 to adapt to that, create symlink, or rename the binary after building
3695 * libudev no longer provides these symbols:
3696 udev_monitor_from_socket()
3697 udev_queue_get_failed_list_entry()
3698 udev_get_{dev,sys,run}_path()
3699 The versions number was bumped and symbol versioning introduced.
3701 * systemd-loginctl and systemd-journalctl have been renamed
3702 to loginctl and journalctl to match systemctl.
3704 * The config files: /etc/systemd/systemd-logind.conf and
3705 /etc/systemd/systemd-journald.conf have been renamed to
3706 logind.conf and journald.conf. Package updates should rename
3707 the files to the new names on upgrade.
3709 * For almost all files the license is now LGPL2.1+, changed
3710 from the previous GPL2.0+. Exceptions are some minor stuff
3711 of udev (which will be changed to LGPL2.1 eventually, too),
3712 and the MIT licensed sd-daemon.[ch] library that is suitable
3713 to be used as drop-in files.
3715 * systemd and logind now handle system sleep states, in
3716 particular suspending and hibernating.
3718 * logind now implements a sleep/shutdown/idle inhibiting logic
3719 suitable for a variety of uses. Soonishly Lennart will blog
3720 about this in more detail.
3722 * var-run.mount and var-lock.mount are no longer provided
3723 (which prevously bind mounted these directories to their new
3724 places). Distributions which have not converted these
3725 directories to symlinks should consider stealing these files
3726 from git history and add them downstream.
3728 * We introduced the Documentation= field for units and added
3729 this to all our shipped units. This is useful to make it
3730 easier to explore the boot and the purpose of the various
3733 * All smaller setup units (such as
3734 systemd-vconsole-setup.service) now detect properly if they
3735 are run in a container and are skipped when
3736 appropriate. This guarantees an entirely noise-free boot in
3737 Linux container environments such as systemd-nspawn.
3739 * A framework for implementing offline system updates is now
3740 integrated, for details see:
3741 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3743 * A new service type Type=idle is available now which helps us
3744 avoiding ugly interleaving of getty output and boot status
3747 * There's now a system-wide CapabilityBoundingSet= option to
3748 globally reduce the set of capabilities for the
3749 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3750 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3751 even CAP_NET_ADMIN system-wide for secure systems.
3753 * There are now system-wide DefaultLimitXXX= options to
3754 globally change the defaults of the various resource limits
3755 for all units started by PID 1.
3757 * Harald Hoyer's systemd test suite has been integrated into
3758 systemd which allows easy testing of systemd builds in qemu
3759 and nspawn. (This is really awesome! Ask us for details!)
3761 * The fstab parser is now implemented as generator, not inside
3764 * systemctl will now warn you if .mount units generated from
3765 /etc/fstab are out of date due to changes in fstab that
3766 have not been read by systemd yet.
3768 * systemd is now suitable for usage in initrds. Dracut has
3769 already been updated to make use of this. With this in place
3770 initrds get a slight bit faster but primarily are much
3771 easier to introspect and debug since "systemctl status" in
3772 the host system can be used to introspect initrd services,
3773 and the journal from the initrd is kept around too.
3775 * systemd-delta has been added, a tool to explore differences
3776 between user/admin configuration and vendor defaults.
3778 * PrivateTmp= now affects both /tmp and /var/tmp.
3780 * Boot time status messages are now much prettier and feature
3781 proper english language. Booting up systemd has never been
3784 * Read-ahead pack files now include the inode number of all
3785 files to pre-cache. When the inode changes the pre-caching
3786 is not attempted. This should be nicer to deal with updated
3787 packages which might result in changes of read-ahead
3790 * We now temporaritly lower the kernel's read_ahead_kb variable
3791 when collecting read-ahead data to ensure the kernel's
3792 built-in read-ahead does not add noise to our measurements
3793 of necessary blocks to pre-cache.
3795 * There's now RequiresMountsFor= to add automatic dependencies
3796 for all mounts necessary for a specific file system path.
3798 * MountAuto= and SwapAuto= have been removed from
3799 system.conf. Mounting file systems at boot has to take place
3802 * nspawn now learned a new switch --uuid= to set the machine
3803 ID on the command line.
3805 * nspawn now learned the -b switch to automatically search
3808 * vt102 is now the default TERM for serial TTYs, upgraded from
3811 * systemd-logind now works on VT-less systems.
3813 * The build tree has been reorganized. The individual
3814 components now have directories of their own.
3816 * A new condition type ConditionPathIsReadWrite= is now available.
3818 * nspawn learned the new -C switch to create cgroups for the
3819 container in other hierarchies.
3821 * We now have support for hardware watchdogs, configurable in
3824 * The scheduled shutdown logic now has a public API.
3826 * We now mount /tmp as tmpfs by default, but this can be
3827 masked and /etc/fstab can override it.
3829 * Since udisks does not make use of /media anymore we are not
3830 mounting a tmpfs on it anymore.
3832 * journalctl gained a new --local switch to only interleave
3833 locally generated journal files.
3835 * We can now load the IMA policy at boot automatically.
3837 * The GTK tools have been split off into a systemd-ui.
3839 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3840 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3841 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3842 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3843 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3844 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3845 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3846 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3847 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3852 * This is mostly a bugfix release
3854 * Support optional initialization of the machine ID from the
3855 KVM or container configured UUID.
3857 * Support immediate reboots with "systemctl reboot -ff"
3859 * Show /etc/os-release data in systemd-analyze output
3861 * Many bugfixes for the journal, including endianness fixes and
3862 ensuring that disk space enforcement works
3864 * sd-login.h is C++ comptaible again
3866 * Extend the /etc/os-release format on request of the Debian
3869 * We now refuse non-UTF8 strings used in various configuration
3870 and unit files. This is done to ensure we do not pass invalid
3871 data over D-Bus or expose it elsewhere.
3873 * Register Mimo USB Screens as suitable for automatic seat
3876 * Read SELinux client context from journal clients in a race
3879 * Reorder configuration file lookup order. /etc now always
3880 overrides /run in order to allow the administrator to always
3881 and unconditionally override vendor supplied or
3882 automatically generated data.
3884 * The various user visible bits of the journal now have man
3885 pages. We still lack man pages for the journal API calls
3888 * We now ship all man pages in HTML format again in the
3891 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3892 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3893 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3898 * This is mostly a bugfix release
3900 * systems lacking /etc/os-release are no longer supported.
3902 * Various functionality updates to libsystemd-login.so
3904 * Track class of PAM logins to distuingish greeters from
3907 Contributions from: Kay Sievers, Lennart Poettering, Michael
3912 * This is an important bugfix release for v41.
3914 * Building man pages is now optional which should be useful
3915 for those building systemd from git but unwilling to install
3918 * Watchdog support for supervising services is now usable. In
3919 a future release support for hardware watchdogs
3920 (i.e. /dev/watchdog) will be added building on this.
3922 * Service start rate limiting is now configurable and can be
3923 turned off per service. When a start rate limit is hit a
3924 reboot can automatically be triggered.
3926 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3928 Contributions from: Benjamin Franzke, Bill Nottingham,
3929 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3930 Schmidt, Michał Górny, Piotr Drąg
3934 * The systemd binary is installed /usr/lib/systemd/systemd now;
3935 An existing /sbin/init symlink needs to be adapted with the
3938 * The code that loads kernel modules has been ported to invoke
3939 libkmod directly, instead of modprobe. This means we do not
3940 support systems with module-init-tools anymore.
3942 * Watchdog support is now already useful, but still not
3945 * A new kernel command line option systemd.setenv= is
3946 understood to set system wide environment variables
3947 dynamically at boot.
3949 * We now limit the set of capabilities of systemd-journald.
3951 * We now set SIGPIPE to ignore by default, since it only is
3952 useful in shell pipelines, and has little use in general
3953 code. This can be disabled with IgnoreSIPIPE=no in unit
3956 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
3957 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
3962 * This is mostly a bugfix release
3964 * We now expose the reason why a service failed in the
3965 "Result" D-Bus property.
3967 * Rudimentary service watchdog support (will be completed over
3968 the next few releases.)
3970 * When systemd forks off in order execute some service we will
3971 now immediately changes its argv[0] to reflect which process
3972 it will execute. This is useful to minimize the time window
3973 with a generic argv[0], which makes bootcharts more useful
3975 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
3976 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
3977 Mike Kazantsev, Ray Strode
3981 * This is mostly a test release, but incorporates many
3984 * New systemd-cgtop tool to show control groups by their
3987 * Linking against libacl for ACLs is optional again. If
3988 disabled, support tracking device access for active logins
3989 goes becomes unavailable, and so does access to the user
3990 journals by the respective users.
3992 * If a group "adm" exists, journal files are automatically
3993 owned by them, thus allow members of this group full access
3994 to the system journal as well as all user journals.
3996 * The journal now stores the SELinux context of the logging
3997 client for all entries.
3999 * Add C++ inclusion guards to all public headers
4001 * New output mode "cat" in the journal to print only text
4002 messages, without any meta data like date or time.
4004 * Include tiny X server wrapper as a temporary stop-gap to
4005 teach XOrg udev display enumeration. This is used by display
4006 managers such as gdm, and will go away as soon as XOrg
4007 learned native udev hotplugging for display devices.
4009 * Add new systemd-cat tool for executing arbitrary programs
4010 with STDERR/STDOUT connected to the journal. Can also act as
4011 BSD logger replacement, and does so by default.
4013 * Optionally store all locally generated coredumps in the
4014 journal along with meta data.
4016 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
4017 writing short strings to files (for usage for /sys), and for
4018 creating symlinks, character and block device nodes.
4020 * New unit file option ControlGroupPersistent= to make cgroups
4021 persistent, following the mechanisms outlined in
4022 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
4024 * Support multiple local RTCs in a sane way
4026 * No longer monopolize IO when replaying readahead data on
4027 rotating disks, since we might starve non-file-system IO to
4028 death, since fanotify() will not see accesses done by blkid,
4031 * Do not show kernel threads in systemd-cgls anymore, unless
4032 requested with new -k switch.
4034 Contributions from: Dan Horák, Kay Sievers, Lennart
4035 Poettering, Michal Schmidt
4039 * This is mostly a test release, but incorporates many
4042 * The git repository moved to:
4043 git://anongit.freedesktop.org/systemd/systemd
4044 ssh://git.freedesktop.org/git/systemd/systemd
4046 * First release with the journal
4047 http://0pointer.de/blog/projects/the-journal.html
4049 * The journal replaces both systemd-kmsg-syslogd and
4050 systemd-stdout-bridge.
4052 * New sd_pid_get_unit() API call in libsystemd-logind
4054 * Many systemadm clean-ups
4056 * Introduce remote-fs-pre.target which is ordered before all
4057 remote mounts and may be used to start services before all
4060 * Added Mageia support
4062 * Add bash completion for systemd-loginctl
4064 * Actively monitor PID file creation for daemons which exit in
4065 the parent process before having finished writing the PID
4066 file in the daemon process. Daemons which do this need to be
4067 fixed (i.e. PID file creation must have finished before the
4068 parent exits), but we now react a bit more gracefully to them.
4070 * Add colourful boot output, mimicking the well-known output
4071 of existing distributions.
4073 * New option PassCredentials= for socket units, for
4074 compatibility with a recent kernel ABI breakage.
4076 * /etc/rc.local is now hooked in via a generator binary, and
4077 thus will no longer act as synchronization point during
4080 * systemctl list-unit-files now supports --root=.
4082 * systemd-tmpfiles now understands two new commands: z, Z for
4083 relabelling files according to the SELinux database. This is
4084 useful to apply SELinux labels to specific files in /sys,
4087 * Output of SysV services is now forwarded to both the console
4088 and the journal by default, not only just the console.
4090 * New man pages for all APIs from libsystemd-login.
4092 * The build tree got reorganized and a the build system is a
4093 lot more modular allowing embedded setups to specifically
4094 select the components of systemd they are interested in.
4096 * Support for Linux systems lacking the kernel VT subsystem is
4099 * configure's --with-rootdir= got renamed to
4100 --with-rootprefix= to follow the naming used by udev and
4103 * Unless specified otherwise we will now install to /usr instead
4104 of /usr/local by default.
4106 * Processes with '@' in argv[0][0] are now excluded from the
4107 final shut-down killing spree, following the logic explained
4109 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
4111 * All processes remaining in a service cgroup when we enter
4112 the START or START_PRE states are now killed with
4113 SIGKILL. That means it is no longer possible to spawn
4114 background processes from ExecStart= lines (which was never
4115 supported anyway, and bad style).
4117 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
4118 reloading of units together.
4120 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
4121 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
4122 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
4123 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
4124 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek