1 systemd System and Service Manager
5 * journalctl gained the new options -t/--identifier= to match
6 on the syslog identifier (aka "tag"), as well as --utc to
7 show log timestamps in the UTC timezone. journalctl now also
8 accepts -n/--lines=all to disable line capping in a pager.
10 * Services can notify the manager before they start a reload
11 (by sending RELOADING=1) or shutdown (by sending
12 STOPPING=1). This allows the manager to track and show the
13 internal state of daemons and closes a race condition when
14 the process is still running but has closed its D-Bus
17 * Services with Type=oneshot do not have to have any ExecStart
20 * User units are now loaded also from
21 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
22 /run/systemd/user directory that was already previously
23 supported, but is under the control of the user.
25 * Job timeouts (i.e. time-outs on the time a job that is
26 queued stays in the run queue) can now optionally result in
27 immediate reboot or power-off actions (JobTimeoutAction= and
28 JobTimeoutRebootArgument=). This is useful on ".target"
29 units, to limit the maximum time a target remains
30 undispatched in the run queue, and to trigger an emergency
31 operation in such a case. This is now used by default to
32 turn off the system if boot-up (as defined by everything in
33 basic.target) hangs and does not complete for at least
34 15min. Also, if power-off or reboot hang for at least 30min
35 an immediate power-off/reboot operation is triggered. This
36 functionality is particularly useful to increase reliability
37 on embedded devices, but also on laptops which might
38 accidentally get powered on when carried in a backpack and
39 whose boot stays stuck in a hard disk encryption passphrase
42 * systemd-logind can be configured to also handle lid switch
43 events even when the machine is docked or multiple displays
44 are attached (HandleLidSwitchDocked= option).
46 * A helper binary and a service have been added which can be
47 used to resume from hibernation in the initramfs. A
48 generator will parse the resume= option on the kernel
49 command-line to trigger resume.
51 * A user console daemon systemd-consoled has been
52 added. Currently, it is a preview, and will so far open a
53 single terminal on each session of the user marked as
54 Desktop=SYSTEMD-CONSOLE.
56 * Route metrics can be specified for DHCP routes added by
59 * The SELinux context of socket-actived services can be set
60 from the information provided by the networking stack
61 (SELinuxContextFromNet= option).
63 * Userspace firmware loading support has been removed and
64 the minimum supported kernel version is thus bumped to 3.7.
66 * Timeout for udev workers has been increased from 1 to 3
67 minutes, but a warning will be printed after 1 minute to
68 help diagnose kernel modules that take a long time to load.
70 * Udev rules can now remove tags on devices with TAG-="foobar".
72 * systemd's readahead implementation has been removed. In many
73 circumstatances it didn't give expected benefits even for
74 rotational disk drives and was becoming less relevant in the
75 age of SSDs. As none of the developers has been using
76 rotating media anymore, and nobody stepped up to actively
77 maintain this component of systemd it has now been removed.
79 * Swap units can use Discard= to specify discard options.
80 Discard options specified for swaps in /etc/fstab are now
83 * Docker containers are now detected as a separate type of
86 * The Password Agent protocol gained support for queries where
87 the user input is shown, useful e.g. for usernames.
88 systemd-ask-password gained a new --echo option to turn that
91 * The default sysctl.d/ snippets will now set:
93 net.core.default_qdisc = fq_codel
95 This selects Fair Queueing Controlled Delay as the default
96 queueing discipline for network interfaces. fq_codel helps
97 fight the network bufferbloat problem. It is believed to be
98 a good default with no tuning required for most workloads.
99 Downstream distributions may override this choice. On 10Gbit
100 servers that do not do forwarding, "fq" may perform better.
101 Systems without a good clocksource should use "pfifo_fast".
103 * If kdbus is enabled during build a new option BusPolicy= is
104 available for service units, that allows locking all service
105 processes into a stricter bus policy, in order to limit
106 access to various bus services, or even hide most of them
107 from the service's view entirely.
109 * networkctl will now show the .network and .link file
110 networkd has applied to a specific interface.
112 * sd-login gained a new API call sd_session_get_desktop() to
113 query which desktop environment has been selected for a
116 * UNIX utmp support is now compile-time optional to support
119 * systemctl gained two new commands "add-wants" and
120 "add-requires" for pulling in units from specific targets
123 * If the word "rescue" is specified on the kernel command line
124 the system will now boot into rescue mode (aka
125 rescue.target), which was previously available only by
126 specifying "1" or "systemd.unit=rescue.target" on the kernel
127 command line. This new kernel command line option nicely
128 mirrors the already existing "emergency" kernel command line
131 * New kernel command line options mount.usr=, mount.usrflags=,
132 usrfstype= have been added that match root=, rootflags=,
133 rootfstype= but allow mounting a specific file system to
136 * The $NOTIFY_SOCKET is now also passed to control processesof
137 services, not only the main process.
139 * This version reenables support for fsck's -l switch. This
140 means at least version v2.25 of util-linux is required for
141 operation, otherwise dead-locks on device nodes may
142 occur. Again: you need to update util-linux to at least
143 v2.25 when updating systemd to v217.
147 * timedated no longer reads NTP implementation unit names from
148 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
149 implementations should add a
151 Conflicts=systemd-timesyncd.service
153 to their unit files to take over and replace systemd's NTP
154 default functionality.
156 * systemd-sysusers gained a new line type "r" for configuring
157 which UID/GID ranges to allocate system users/groups
158 from. Lines of type "u" may now add an additional column
159 that specifies the home directory for the system user to be
160 created. Also, systemd-sysusers may now optionally read user
161 information from STDIN instead of a file. This is useful for
162 invoking it from RPM preinst scriptlets that need to create
163 users before the first RPM file is installed since these
164 files might need to be owned by them. A new
165 %sysusers_create_inline RPM macro has been introduced to do
166 just that. systemd-sysusers now updates the shadow files as
167 well as the user/group databases, which should enhance
168 compatibility with certain tools like grpck.
170 * A number of bus APIs of PID 1 now optionally consult
171 PolicyKit to permit access for otherwise unprivileged
172 clients under certain conditions. Note that this currently
173 doesn't support interactive authentication yet, but this is
174 expected to be added eventually, too.
176 * /etc/machine-info now has new fields for configuring the
177 deployment environment of the machine, as well as the
178 location of the machine. hostnamectl has been updated with
179 new command to update these fields.
181 * systemd-timesyncd has been updated to automatically acquire
182 NTP server information from systemd-networkd, which might
183 have been discovered via DHCP.
185 * systemd-resolved now includes a caching DNS stub resolver
186 and a complete LLMNR name resolution implementation. A new
187 NSS module "nss-resolve" has been added which can be used
188 instead of glibc's own "nss-dns" to resolve hostnames via
189 systemd-resolved. Hostnames, addresses and arbitrary RRs may
190 be resolved via systemd-resolved D-Bus APIs. In contrast to
191 the glibc internal resolver systemd-resolved is aware of
192 multi-homed system, and keeps DNS server and caches separate
193 and per-interface. Queries are sent simultaneously on all
194 interfaces that have DNS servers configured, in order to
195 properly handle VPNs and local LANs which might resolve
196 separate sets of domain names. systemd-resolved may acquire
197 DNS server information from systemd-networkd automatically,
198 which in turn might have discovered them via DHCP. A tool
199 "systemd-resolve-host" has been added that may be used to
200 query the DNS logic in resolved. systemd-resolved implements
201 IDNA and automatically uses IDNA or UTF-8 encoding depending
202 on whether classic DNS or LLMNR is used as transport. In the
203 next releases we intend to add a DNSSEC and mDNS/DNS-SD
204 implementation to systemd-resolved.
206 * A new NSS module nss-mymachines has been added, that
207 automatically resolves the names of all local registered
208 containers to their respective IP addresses.
210 * A new client tool "networkctl" for systemd-networkd has been
211 added. It currently is entirely passive and will query
212 networking configuration from udev, rtnetlink and networkd,
213 and present it to the user in a very friendly
214 way. Eventually, we hope to extend it to become a full
215 control utility for networkd.
217 * .socket units gained a new DeferAcceptSec= setting that
218 controls the kernels' TCP_DEFER_ACCEPT sockopt for
219 TCP. Similar, support for controlling TCP keep-alive
220 settings has been added (KeepAliveTimeSec=,
221 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
222 turning off Nagle's algorithm on TCP has been added
225 * logind learned a new session type "web", for use in projects
226 like Cockpit which register web clients as PAM sessions.
228 * timer units with at least one OnCalendar= setting will now
229 be started only after timer-sync.target has been
230 reached. This way they will not elapse before the system
231 clock has been corrected by a local NTP client or
232 similar. This is particular useful on RTC-less embedded
233 machines, that come up with an invalid system clock.
235 * systemd-nspawn's --network-veth= switch should now result in
236 stable MAC addresses for both the outer and the inner side
239 * systemd-nspawn gained a new --volatile= switch for running
240 container instances with /etc or /var unpopulated.
242 * The kdbus client code has been updated to use the new Linux
243 3.17 memfd subsystem instead of the old kdbus-specific one.
245 * systemd-networkd's DHCP client and server now support
246 FORCERENEW. There are also new configuration options to
247 configure the vendor client identifier and broadcast mode
250 * systemd will no longer inform the kernel about the current
251 timezone, as this is necessarily incorrect and racy as the
252 kernel has no understanding of DST and similar
253 concepts. This hence means FAT timestamps will be always
254 considered UTC, similar to what Android is already
255 doing. Also, when the RTC is configured to the local time
256 (rather than UTC) systemd will never synchronize back to it,
257 as this might confuse Windows at a later boot.
259 * systemd-analyze gained a new command "verify" for offline
260 validation of unit files.
262 * systemd-networkd gained support for a couple of additional
263 settings for bonding networking setups. Also, the metric for
264 statically configured routes may now be configured. For
265 network interfaces where this is appropriate the peer IP
266 address may now be configured.
268 * systemd-networkd's DHCP client will no longer request
269 broadcasting by default, as this tripped up some networks.
270 For hardware where broadcast is required the feature should
271 be switched back on using RequestBroadcast=yes.
273 * systemd-networkd will now set up IPv4LL addresses (when
274 enabled) even if DHCP is configured successfully.
276 * udev will now default to respect network device names given
277 by the kernel when the kernel indicates that these are
278 predictable. This behavior can be tweaked by changing
279 NamePolicy= in the relevant .link file.
281 * A new library systemd-terminal has been added that
282 implements full TTY stream parsing and rendering. This
283 library is supposed to be used later on for implementing a
284 full userspace VT subsystem, replacing the current kernel
287 * A new tool systemd-journal-upload has been added to push
288 journal data to a remote system running
289 systemd-journal-remote.
291 * journald will no longer forward all local data to another
292 running syslog daemon. This change has been made because
293 rsyslog (which appears to be the most commonly used syslog
294 implementation these days) no longer makes use of this, and
295 instead pulls the data out of the journal on its own. Since
296 forwarding the messages to a non-existent syslog server is
297 more expensive than we assumed we have now turned this
298 off. If you run a syslog server that is not a recent rsyslog
299 version, you have to turn this option on again
300 (ForwardToSyslog= in journald.conf).
302 * journald now optionally supports the LZ4 compressor for
303 larger journal fields. This compressor should perform much
304 better than XZ which was the previous default.
306 * machinectl now shows the IP addresses of local containers,
307 if it knows them, plus the interface name of the container.
309 * A new tool "systemd-escape" has been added that makes it
310 easy to escape strings to build unit names and similar.
312 * sd_notify() messages may now include a new ERRNO= field
313 which is parsed and collected by systemd and shown among the
314 "systemctl status" output for a service.
316 * A new component "systemd-firstboot" has been added that
317 queries the most basic systemd information (timezone,
318 hostname, root password) interactively on first
319 boot. Alternatively it may also be used to provision these
320 things offline on OS images installed into directories.
322 * The default sysctl.d/ snippets will now set
324 net.ipv4.conf.default.promote_secondaries=1
326 This has the benefit of no flushing secondary IP addresses
327 when primary addresses are removed.
329 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
330 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
331 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
332 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
333 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
334 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
335 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
336 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
337 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
338 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
339 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
340 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
341 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
342 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
343 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
345 -- Berlin, 2014-08-19
349 * A new tool systemd-sysusers has been added. This tool
350 creates system users and groups in /etc/passwd and
351 /etc/group, based on static declarative system user/group
352 definitions in /usr/lib/sysusers.d/. This is useful to
353 enable factory resets and volatile systems that boot up with
354 an empty /etc directory, and thus need system users and
355 groups created during early boot. systemd now also ships
356 with two default sysusers.d/ files for the most basic
357 users and groups systemd and the core operating system
360 * A new tmpfiles snippet has been added that rebuilds the
361 essential files in /etc on boot, should they be missing.
363 * A directive for ensuring automatic clean-up of
364 /var/cache/man/ has been removed from the default
365 configuration. This line should now be shipped by the man
366 implementation. The necessary change has been made to the
367 man-db implementation. Note that you need to update your man
368 implementation to one that ships this line, otherwise no
369 automatic clean-up of /var/cache/man will take place.
371 * A new condition ConditionNeedsUpdate= has been added that
372 may conditionalize services to only run when /etc or /var
373 are "older" than the vendor operating system resources in
374 /usr. This is useful for reconstructing or updating /etc
375 after an offline update of /usr or a factory reset, on the
376 next reboot. Services that want to run once after such an
377 update or reset should use this condition and order
378 themselves before the new systemd-update-done.service, which
379 will mark the two directories as fully updated. A number of
380 service files have been added making use of this, to rebuild
381 the udev hardware database, the journald message catalog and
382 dynamic loader cache (ldconfig). The systemd-sysusers tool
383 described above also makes use of this now. With this in
384 place it is now possible to start up a minimal operating
385 system with /etc empty cleanly. For more information on the
386 concepts involved see this recent blog story:
388 http://0pointer.de/blog/projects/stateless.html
390 * A new system group "input" has been introduced, and all
391 input device nodes get this group assigned. This is useful
392 for system-level software to get access to input devices. It
393 complements what is already done for "audio" and "video".
395 * systemd-networkd learnt minimal DHCPv4 server support in
396 addition to the existing DHCPv4 client support. It also
397 learnt DHCPv6 client and IPv6 Router Solicitation client
398 support. The DHCPv4 client gained support for static routes
399 passed in from the server. Note that the [DHCPv4] section
400 known in older systemd-networkd versions has been renamed to
401 [DHCP] and is now also used by the DHCPv6 client. Existing
402 .network files using settings of this section should be
403 updated, though compatibility is maintained. Optionally, the
404 client hostname may now be sent to the DHCP server.
406 * networkd gained support for vxlan virtual networks as well
407 as tun/tap and dummy devices.
409 * networkd gained support for automatic allocation of address
410 ranges for interfaces from a system-wide pool of
411 addresses. This is useful for dynamically managing a large
412 number of interfaces with a single network configuration
413 file. In particular this is useful to easily assign
414 appropriate IP addresses to the veth links of a large number
417 * RPM macros for processing sysusers, sysctl and binfmt
418 drop-in snippets at package installation time have been
421 * The /etc/os-release file should now be placed in
422 /usr/lib/os-release. The old location is automatically
423 created as symlink. /usr/lib is the more appropriate
424 location of this file, since it shall actually describe the
425 vendor operating system shipped in /usr, and not the
426 configuration stored in /etc.
428 * .mount units gained a new boolean SloppyOptions= setting
429 that maps to mount(8)'s -s option which enables permissive
430 parsing of unknown mount options.
432 * tmpfiles learnt a new "L+" directive which creates a symlink
433 but (unlike "L") deletes a pre-existing file first, should
434 it already exist and not already be the correct
435 symlink. Similar, "b+", "c+" and "p+" directives have been
436 added as well, which create block and character devices, as
437 well as fifos in the filesystem, possibly removing any
438 pre-existing files of different types.
440 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
441 'argument' field (which so far specified the source to
442 symlink/copy the files from) is now optional. If omitted the
443 same file os copied from /usr/share/factory/ suffixed by the
444 full destination path. This is useful for populating /etc
445 with essential files, by copying them from vendor defaults
446 shipped in /usr/share/factory/etc.
448 * A new command "systemctl preset-all" has been added that
449 applies the service preset settings to all installed unit
450 files. A new switch --preset-mode= has been added that
451 controls whether only enable or only disable operations
454 * A new command "systemctl is-system-running" has been added
455 that allows checking the overall state of the system, for
456 example whether it is fully up and running.
458 * When the system boots up with an empty /etc, the equivalent
459 to "systemctl preset-all" is executed during early boot, to
460 make sure all default services are enabled after a factory
463 * systemd now contains a minimal preset file that enables the
464 most basic services systemd ships by default.
466 * Unit files' [Install] section gained a new DefaultInstance=
467 field for defining the default instance to create if a
468 template unit is enabled with no instance specified.
470 * A new passive target cryptsetup-pre.target has been added
471 that may be used by services that need to make they run and
472 finish before the first LUKS cryptographic device is set up.
474 * The /dev/loop-control and /dev/btrfs-control device nodes
475 are now owned by the "disk" group by default, opening up
476 access to this group.
478 * systemd-coredump will now automatically generate a
479 stack trace of all core dumps taking place on the system,
480 based on elfutils' libdw library. This stack trace is logged
483 * systemd-coredump may now optionally store coredumps directly
484 on disk (in /var/lib/systemd/coredump, possibly compressed),
485 instead of storing them unconditionally in the journal. This
486 mode is the new default. A new configuration file
487 /etc/systemd/coredump.conf has been added to configure this
488 and other parameters of systemd-coredump.
490 * coredumpctl gained a new "info" verb to show details about a
491 specific coredump. A new switch "-1" has also been added
492 that makes sure to only show information about the most
493 recent entry instead of all entries. Also, as the tool is
494 generally useful now the "systemd-" prefix of the binary
495 name has been removed. Distributions that want to maintain
496 compatibility with the old name should add a symlink from
497 the old name to the new name.
499 * journald's SplitMode= now defaults to "uid". This makes sure
500 that unprivileged users can access their own coredumps with
501 coredumpctl without restrictions.
503 * New kernel command line options "systemd.wants=" (for
504 pulling an additional unit during boot), "systemd.mask="
505 (for masking a specific unit for the boot), and
506 "systemd.debug-shell" (for enabling the debug shell on tty9)
507 have been added. This is implemented in the new generator
508 "systemd-debug-generator".
510 * systemd-nspawn will now by default filter a couple of
511 syscalls for containers, among them those required for
512 kernel module loading, direct x86 IO port access, swap
513 management, and kexec. Most importantly though
514 open_by_handle_at() is now prohibited for containers,
515 closing a hole similar to a recently discussed vulnerability
516 in docker regarding access to files on file hierarchies the
517 container should normally not have access to. Note that for
518 nspawn we generally make no security claims anyway (and
519 this is explicitly documented in the man page), so this is
520 just a fix for one of the most obvious problems.
522 * A new man page file-hierarchy(7) has been added that
523 contains a minimized, modernized version of the file system
524 layout systemd expects, similar in style to the FHS
525 specification or hier(5). A new tool systemd-path(1) has
526 been added to query many of these paths for the local
529 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
530 longer done. Since the directory now has a per-user size
531 limit, and is cleaned on logout this appears unnecessary,
532 in particular since this now brings the lifecycle of this
533 directory closer in line with how IPC objects are handled.
535 * systemd.pc now exports a number of additional directories,
536 including $libdir (which is useful to identify the library
537 path for the primary architecture of the system), and a
538 couple of drop-in directories.
540 * udev's predictable network interface names now use the dev_port
541 sysfs attribute, introduced in linux 3.15 instead of dev_id to
542 distinguish between ports of the same PCI function. dev_id should
543 only be used for ports using the same HW address, hence the need
546 * machined has been updated to export the OS version of a
547 container (read from /etc/os-release and
548 /usr/lib/os-release) on the bus. This is now shown in
549 "machinectl status" for a machine.
551 * A new service setting RestartForceExitStatus= has been
552 added. If configured to a set of exit signals or process
553 return values, the service will be restarted when the main
554 daemon process exits with any of them, regardless of the
557 * systemctl's -H switch for connecting to remote systemd
558 machines has been extended so that it may be used to
559 directly connect to a specific container on the
560 host. "systemctl -H root@foobar:waldi" will now connect as
561 user "root" to host "foobar", and then proceed directly to
562 the container named "waldi". Note that currently you have to
563 authenticate as user "root" for this to work, as entering
564 containers is a privileged operation.
566 Contributions from: Andreas Henriksson, Benjamin Steinwender,
567 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
568 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
569 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
570 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
571 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
572 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
573 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
574 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
575 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
576 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
577 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
579 -- Berlin, 2014-07-03
583 * As an experimental feature, udev now tries to lock the
584 disk device node (flock(LOCK_SH|LOCK_NB)) while it
585 executes events for the disk or any of its partitions.
586 Applications like partitioning programs can lock the
587 disk device node (flock(LOCK_EX)) and claim temporary
588 device ownership that way; udev will entirely skip all event
589 handling for this disk and its partitions. If the disk
590 was opened for writing, the close will trigger a partition
591 table rescan in udev's "watch" facility, and if needed
592 synthesize "change" events for the disk and all its partitions.
593 This is now unconditionally enabled, and if it turns out to
594 cause major problems, we might turn it on only for specific
595 devices, or might need to disable it entirely. Device Mapper
596 devices are excluded from this logic.
598 * We temporarily dropped the "-l" switch for fsck invocations,
599 since they collide with the flock() logic above. util-linux
600 upstream has been changed already to avoid this conflict,
601 and we will readd "-l" as soon as util-linux with this
602 change has been released.
604 * The dependency on libattr has been removed. Since a long
605 time, the extended attribute calls have moved to glibc, and
606 libattr is thus unnecessary.
608 * Virtualization detection works without priviliges now. This
609 means the systemd-detect-virt binary no longer requires
610 CAP_SYS_PTRACE file capabilities, and our daemons can run
611 with fewer privileges.
613 * systemd-networkd now runs under its own "systemd-network"
614 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
615 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
616 loses the ability to write to files owned by root this way.
618 * Similar, systemd-resolved now runs under its own
619 "systemd-resolve" user with no capabilities remaining.
621 * Similar, systemd-bus-proxyd now runs under its own
622 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
624 * systemd-networkd gained support for setting up "veth"
625 virtual ethernet devices for container connectivity, as well
626 as GRE and VTI tunnels.
628 * systemd-networkd will no longer automatically attempt to
629 manually load kernel modules necessary for certain tunnel
630 transports. Instead, it is assumed the kernel loads them
631 automatically when required. This only works correctly on
632 very new kernels. On older kernels, please consider adding
633 the kernel modules to /etc/modules-load.d/ as a work-around.
635 * The resolv.conf file systemd-resolved generates has been
636 moved to /run/systemd/resolve/. If you have a symlink from
637 /etc/resolv.conf, it might be necessary to correct it.
639 * Two new service settings, ProtectHome= and ProtectSystem=,
640 have been added. When enabled, they will make the user data
641 (such as /home) inaccessible or read-only and the system
642 (such as /usr) read-only, for specific services. This allows
643 very light-weight per-service sandboxing to avoid
644 modifications of user data or system files from
645 services. These two new switches have been enabled for all
646 of systemd's long-running services, where appropriate.
648 * Socket units gained new SocketUser= and SocketGroup=
649 settings to set the owner user and group of AF_UNIX sockets
650 and FIFOs in the file system.
652 * Socket units gained a new RemoveOnStop= setting. If enabled,
653 all FIFOS and sockets in the file system will be removed
654 when the specific socket unit is stopped.
656 * Socket units gained a new Symlinks= setting. It takes a list
657 of symlinks to create to file system sockets or FIFOs
658 created by the specific Unix sockets. This is useful to
659 manage symlinks to socket nodes with the same life-cycle as
662 * The /dev/log socket and /dev/initctl FIFO have been moved to
663 /run, and have been replaced by symlinks. This allows
664 connecting to these facilities even if PrivateDevices=yes is
665 used for a service (which makes /dev/log itself unavailable,
666 but /run is left). This also has the benefit of ensuring
667 that /dev only contains device nodes, directories and
668 symlinks, and nothing else.
670 * sd-daemon gained two new calls sd_pid_notify() and
671 sd_pid_notifyf(). They are similar to sd_notify() and
672 sd_notifyf(), but allow overriding of the source PID of
673 notification messages if permissions permit this. This is
674 useful to send notify messages on behalf of a different
675 process (for example, the parent process). The
676 systemd-notify tool has been updated to make use of this
677 when sending messages (so that notification messages now
678 originate from the shell script invoking systemd-notify and
679 not the systemd-notify process itself. This should minimize
680 a race where systemd fails to associate notification
681 messages to services when the originating process already
684 * A new "on-abnormal" setting for Restart= has been added. If
685 set, it will result in automatic restarts on all "abnormal"
686 reasons for a process to exit, which includes unclean
687 signals, core dumps, timeouts and watchdog timeouts, but
688 does not include clean and unclean exit codes or clean
689 signals. Restart=on-abnormal is an alternative for
690 Restart=on-failure for services that shall be able to
691 terminate and avoid restarts on certain errors, by
692 indicating so with an unclean exit code. Restart=on-failure
693 or Restart=on-abnormal is now the recommended setting for
694 all long-running services.
696 * If the InaccessibleDirectories= service setting points to a
697 mount point (or if there are any submounts contained within
698 it), it is now attempted to completely unmount it, to make
699 the file systems truly unavailable for the respective
702 * The ReadOnlyDirectories= service setting and
703 systemd-nspawn's --read-only parameter are now recursively
704 applied to all submounts, too.
706 * Mount units may now be created transiently via the bus APIs.
708 * The support for SysV and LSB init scripts has been removed
709 from the systemd daemon itself. Instead, it is now
710 implemented as a generator that creates native systemd units
711 from these scripts when needed. This enables us to remove a
712 substantial amount of legacy code from PID 1, following the
713 fact that many distributions only ship a very small number
714 of LSB/SysV init scripts nowadays.
716 * Priviliged Xen (dom0) domains are not considered
717 virtualization anymore by the virtualization detection
718 logic. After all, they generally have unrestricted access to
719 the hardware and usually are used to manage the unprivileged
722 * systemd-tmpfiles gained a new "C" line type, for copying
723 files or entire directories.
725 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
726 lines. So far, they have been non-globbing versions of the
727 latter, and have thus been redundant. In future, it is
728 recommended to only use "z". "m" has hence been removed
729 from the documentation, even though it stays supported.
731 * A tmpfiles snippet to recreate the most basic structure in
732 /var has been added. This is enough to create the /var/run →
733 /run symlink and create a couple of structural
734 directories. This allows systems to boot up with an empty or
735 volatile /var. Of course, while with this change, the core OS
736 now is capable with dealing with a volatile /var, not all
737 user services are ready for it. However, we hope that sooner
738 or later, many service daemons will be changed upstream so
739 that they are able to automatically create their necessary
740 directories in /var at boot, should they be missing. This is
741 the first step to allow state-less systems that only require
742 the vendor image for /usr to boot.
744 * systemd-nspawn has gained a new --tmpfs= switch to mount an
745 empty tmpfs instance to a specific directory. This is
746 particularly useful for making use of the automatic
747 reconstruction of /var (see above), by passing --tmpfs=/var.
749 * Access modes specified in tmpfiles snippets may now be
750 prefixed with "~", which indicates that they shall be masked
751 by whether the existing file or directory is currently
752 writable, readable or executable at all. Also, if specified,
753 the sgid/suid/sticky bits will be masked for all
756 * A new passive target unit "network-pre.target" has been
757 added which is useful for services that shall run before any
758 network is configured, for example firewall scripts.
760 * The "floppy" group that previously owned the /dev/fd*
761 devices is no longer used. The "disk" group is now used
762 instead. Distributions should probably deprecate usage of
765 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
766 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
767 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
768 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
769 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
770 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
773 -- Berlin, 2014-06-11
777 * A new "systemd-timesyncd" daemon has been added for
778 synchronizing the system clock across the network. It
779 implements an SNTP client. In contrast to NTP
780 implementations such as chrony or the NTP reference server,
781 this only implements a client side, and does not bother with
782 the full NTP complexity, focusing only on querying time from
783 one remote server and synchronizing the local clock to
784 it. Unless you intend to serve NTP to networked clients or
785 want to connect to local hardware clocks, this simple NTP
786 client should be more than appropriate for most
787 installations. The daemon runs with minimal privileges, and
788 has been hooked up with networkd to only operate when
789 network connectivity is available. The daemon saves the
790 current clock to disk every time a new NTP sync has been
791 acquired, and uses this to possibly correct the system clock
792 early at bootup, in order to accommodate for systems that
793 lack an RTC such as the Raspberry Pi and embedded devices,
794 and to make sure that time monotonically progresses on these
795 systems, even if it is not always correct. To make use of
796 this daemon, a new system user and group "systemd-timesync"
797 needs to be created on installation of systemd.
799 * The queue "seqnum" interface of libudev has been disabled, as
800 it was generally incompatible with device namespacing as
801 sequence numbers of devices go "missing" if the devices are
802 part of a different namespace.
804 * "systemctl list-timers" and "systemctl list-sockets" gained
805 a --recursive switch for showing units of these types also
806 for all local containers, similar in style to the already
807 supported --recursive switch for "systemctl list-units".
809 * A new RebootArgument= setting has been added for service
810 units, which may be used to specify a kernel reboot argument
811 to use when triggering reboots with StartLimitAction=.
813 * A new FailureAction= setting has been added for service
814 units which may be used to specify an operation to trigger
815 when a service fails. This works similarly to
816 StartLimitAction=, but unlike it, controls what is done
817 immediately rather than only after several attempts to
818 restart the service in question.
820 * hostnamed got updated to also expose the kernel name,
821 release, and version on the bus. This is useful for
822 executing commands like hostnamectl with the -H switch.
823 systemd-analyze makes use of this to properly display
824 details when running non-locally.
826 * The bootchart tool can now show cgroup information in the
829 * The CFS CPU quota cgroup attribute is now exposed for
830 services. The new CPUQuota= switch has been added for this
831 which takes a percentage value. Setting this will have the
832 result that a service may never get more CPU time than the
833 specified percentage, even if the machine is otherwise idle.
835 * systemd-networkd learned IPIP and SIT tunnel support.
837 * LSB init scripts exposing a dependency on $network will now
838 get a dependency on network-online.target rather than simply
839 network.target. This should bring LSB handling closer to
840 what it was on SysV systems.
842 * A new fsck.repair= kernel option has been added to control
843 how fsck shall deal with unclean file systems at boot.
845 * The (.ini) configuration file parser will now silently
846 ignore sections whose name begins with "X-". This may be
847 used to maintain application-specific extension sections in unit
850 * machined gained a new API to query the IP addresses of
851 registered containers. "machinectl status" has been updated
852 to show these addresses in its output.
854 * A new call sd_uid_get_display() has been added to the
855 sd-login APIs for querying the "primary" session of a
856 user. The "primary" session of the user is elected from the
857 user's sessions and generally a graphical session is
858 preferred over a text one.
860 * A minimal systemd-resolved daemon has been added. It
861 currently simply acts as a companion to systemd-networkd and
862 manages resolv.conf based on per-interface DNS
863 configuration, possibly supplied via DHCP. In the long run
864 we hope to extend this into a local DNSSEC enabled DNS and
867 * The systemd-networkd-wait-online tool is now enabled by
868 default. It will delay network-online.target until a network
869 connection has been configured. The tool primarily integrates
870 with networkd, but will also make a best effort to make sense
871 of network configuration performed in some other way.
873 * Two new service options StartupCPUShares= and
874 StartupBlockIOWeight= have been added that work similarly to
875 CPUShares= and BlockIOWeight= however only apply during
876 system startup. This is useful to prioritize certain services
877 differently during bootup than during normal runtime.
879 * hostnamed has been changed to prefer the statically
880 configured hostname in /etc/hostname (unless set to
881 'localhost' or empty) over any dynamic one supplied by
882 dhcp. With this change, the rules for picking the hostname
883 match more closely the rules of other configuration settings
884 where the local administrator's configuration in /etc always
885 overrides any other settings.
887 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
888 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
889 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
890 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
891 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
892 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
893 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
894 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
895 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
896 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
897 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
898 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
899 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
900 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
901 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
902 Lindskog, WaLyong Cho, Will Woods, Zbigniew
905 -- Beijing, 2014-05-28
909 * When restoring the screen brightness at boot, stay away from
910 the darkest setting or from the lowest 5% of the available
911 range, depending on which is the larger value of both. This
912 should effectively protect the user from rebooting into a
913 black screen, should the brightness have been set to minimum
916 * sd-login gained a new sd_machine_get_class() call to
917 determine the class ("vm" or "container") of a machine
918 registered with machined.
920 * sd-login gained new calls
921 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
922 to query the identity of the peer of a local AF_UNIX
923 connection. They operate similarly to their sd_pid_get_xyz()
926 * PID 1 will now maintain a system-wide system state engine
927 with the states "starting", "running", "degraded",
928 "maintenance", "stopping". These states are bound to system
929 startup, normal runtime, runtime with at least one failed
930 service, rescue/emergency mode and system shutdown. This
931 state is shown in the "systemctl status" output when no unit
932 name is passed. It is useful to determine system state, in
933 particularly when doing so for many systems or containers at
936 * A new command "list-machines" has been added to "systemctl"
937 that lists all local OS containers and shows their system
938 state (see above), if systemd runs inside of them.
940 * systemctl gained a new "-r" switch to recursively enumerate
941 units on all local containers, when used with the
942 "list-unit" command (which is the default one that is
943 executed when no parameters are specified).
945 * The GPT automatic partition discovery logic will now honour
946 two GPT partition flags: one may be set on a partition to
947 cause it to be mounted read-only, and the other may be set
948 on a partition to ignore it during automatic discovery.
950 * Two new GPT type UUIDs have been added for automatic root
951 partition discovery, for 32-bit and 64-bit ARM. This is not
952 particularly useful for discovering the root directory on
953 these architectures during bare-metal boots (since UEFI is
954 not common there), but still very useful to allow booting of
955 ARM disk images in nspawn with the -i option.
957 * MAC addresses of interfaces created with nspawn's
958 --network-interface= switch will now be generated from the
959 machine name, and thus be stable between multiple invocations
962 * logind will now automatically remove all IPC objects owned
963 by a user if she or he fully logs out. This makes sure that
964 users who are logged out cannot continue to consume IPC
965 resources. This covers SysV memory, semaphores and message
966 queues as well as POSIX shared memory and message
967 queues. Traditionally, SysV and POSIX IPC had no life-cycle
968 limits. With this functionality, that is corrected. This may
969 be turned off by using the RemoveIPC= switch of logind.conf.
971 * The systemd-machine-id-setup and tmpfiles tools gained a
972 --root= switch to operate on a specific root directory,
975 * journald can now forward logged messages to the TTYs of all
976 logged in users ("wall"). This is the default for all
977 emergency messages now.
979 * A new tool systemd-journal-remote has been added to stream
980 journal log messages across the network.
982 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
983 controller trees are mounted into it. Note that the
984 directories mounted beneath it are not read-only. This is a
985 security measure and is particularly useful because glibc
986 actually includes a search logic to pick any tmpfs it can
987 find to implement shm_open() if /dev/shm is not available
988 (which it might very well be in namespaced setups).
990 * machinectl gained a new "poweroff" command to cleanly power
991 down a local OS container.
993 * The PrivateDevices= unit file setting will now also drop the
994 CAP_MKNOD capability from the capability bound set, and
995 imply DevicePolicy=closed.
997 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
998 comprehensively on all long-running systemd services where
1001 * systemd-udevd will now run in a disassociated mount
1002 namespace. To mount directories from udev rules, make sure to
1003 pull in mount units via SYSTEMD_WANTS properties.
1005 * The kdbus support gained support for uploading policy into
1006 the kernel. sd-bus gained support for creating "monitoring"
1007 connections that can eavesdrop into all bus communication
1008 for debugging purposes.
1010 * Timestamps may now be specified in seconds since the UNIX
1011 epoch Jan 1st, 1970 by specifying "@" followed by the value
1014 * Native tcpwrap support in systemd has been removed. tcpwrap
1015 is old code, not really maintained anymore and has serious
1016 shortcomings, and better options such as firewalls
1017 exist. For setups that require tcpwrap usage, please
1018 consider invoking your socket-activated service via tcpd,
1019 like on traditional inetd.
1021 * A new system.conf configuration option
1022 DefaultTimerAccuracySec= has been added that controls the
1023 default AccuracySec= setting of .timer units.
1025 * Timer units gained a new WakeSystem= switch. If enabled,
1026 timers configured this way will cause the system to resume
1027 from system suspend (if the system supports that, which most
1030 * Timer units gained a new Persistent= switch. If enabled,
1031 timers configured this way will save to disk when they have
1032 been last triggered. This information is then used on next
1033 reboot to possible execute overdue timer events, that
1034 could not take place because the system was powered off.
1035 This enables simple anacron-like behaviour for timer units.
1037 * systemctl's "list-timers" will now also list the time a
1038 timer unit was last triggered in addition to the next time
1039 it will be triggered.
1041 * systemd-networkd will now assign predictable IPv4LL
1042 addresses to its local interfaces.
1044 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
1045 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
1046 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
1047 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
1048 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
1049 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
1050 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
1051 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
1054 -- Berlin, 2014-03-25
1058 * A new unit file setting RestrictAddressFamilies= has been
1059 added to restrict which socket address families unit
1060 processes gain access to. This takes address family names
1061 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1062 attack surface of services via exotic protocol stacks. This
1063 is built on seccomp system call filters.
1065 * Two new unit file settings RuntimeDirectory= and
1066 RuntimeDirectoryMode= have been added that may be used to
1067 manage a per-daemon runtime directories below /run. This is
1068 an alternative for setting up directory permissions with
1069 tmpfiles snippets, and has the advantage that the runtime
1070 directory's lifetime is bound to the daemon runtime and that
1071 the daemon starts up with an empty directory each time. This
1072 is particularly useful when writing services that drop
1073 privileges using the User= or Group= setting.
1075 * The DeviceAllow= unit setting now supports globbing for
1076 matching against device group names.
1078 * The systemd configuration file system.conf gained new
1079 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1080 DefaultMemoryAccounting= to globally turn on/off accounting
1081 for specific resources (cgroups) for all units. These
1082 settings may still be overridden individually in each unit
1085 * systemd-gpt-auto-generator is now able to discover /srv and
1086 root partitions in addition to /home and swap partitions. It
1087 also supports LUKS-encrypted partitions now. With this in
1088 place, automatic discovery of partitions to mount following
1089 the Discoverable Partitions Specification
1090 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1091 is now a lot more complete. This allows booting without
1092 /etc/fstab and without root= on the kernel command line on
1093 systems prepared appropriately.
1095 * systemd-nspawn gained a new --image= switch which allows
1096 booting up disk images and Linux installations on any block
1097 device that follow the Discoverable Partitions Specification
1098 (see above). This means that installations made with
1099 appropriately updated installers may now be started and
1100 deployed using container managers, completely
1101 unmodified. (We hope that libvirt-lxc will add support for
1102 this feature soon, too.)
1104 * systemd-nspawn gained a new --network-macvlan= setting to
1105 set up a private macvlan interface for the
1106 container. Similarly, systemd-networkd gained a new
1107 Kind=macvlan setting in .netdev files.
1109 * systemd-networkd now supports configuring local addresses
1112 * A new tool systemd-network-wait-online has been added to
1113 synchronously wait for network connectivity using
1116 * The sd-bus.h bus API gained a new sd_bus_track object for
1117 tracking the life-cycle of bus peers. Note that sd-bus.h is
1118 still not a public API though (unless you specify
1119 --enable-kdbus on the configure command line, which however
1120 voids your warranty and you get no API stability guarantee).
1122 * The $XDG_RUNTIME_DIR runtime directories for each user are
1123 now individual tmpfs instances, which has the benefit of
1124 introducing separate pools for each user, with individual
1125 size limits, and thus making sure that unprivileged clients
1126 can no longer negatively impact the system or other users by
1127 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
1128 RuntimeDirectorySize= has been introduced that allows
1129 controlling the default size limit for all users. It
1130 defaults to 10% of the available physical memory. This is no
1131 replacement for quotas on tmpfs though (which the kernel
1132 still does not support), as /dev/shm and /tmp are still
1133 shared resources used by both the system and unprivileged
1136 * logind will now automatically turn off automatic suspending
1137 on laptop lid close when more than one display is
1138 connected. This was previously expected to be implemented
1139 individually in desktop environments (such as GNOME),
1140 however has been added to logind now, in order to fix a
1141 boot-time race where a desktop environment might not have
1142 been started yet and thus not been able to take an inhibitor
1143 lock at the time where logind already suspends the system
1144 due to a closed lid.
1146 * logind will now wait at least 30s after each system
1147 suspend/resume cycle, and 3min after system boot before
1148 suspending the system due to a closed laptop lid. This
1149 should give USB docking stations and similar enough time to
1150 be probed and configured after system resume and boot in
1151 order to then act as suspend blocker.
1153 * systemd-run gained a new --property= setting which allows
1154 initialization of resource control properties (and others)
1155 for the created scope or service unit. Example: "systemd-run
1156 --property=BlockIOWeight=10 updatedb" may be used to run
1157 updatedb at a low block IO scheduling weight.
1159 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1160 now also work in --scope mode.
1162 * When systemd is compiled with kdbus support, basic support
1163 for enforced policies is now in place. (Note that enabling
1164 kdbus still voids your warranty and no API compatibility
1167 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1168 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1169 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1170 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1171 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1172 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1173 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1174 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
1175 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
1176 Zbigniew Jędrzejewski-Szmek
1178 -- Berlin, 2014-03-12
1182 * systemd will now relabel /dev after loading the SMACK policy
1183 according to SMACK rules.
1185 * A new unit file option AppArmorProfile= has been added to
1186 set the AppArmor profile for the processes of a unit.
1188 * A new condition check ConditionArchitecture= has been added
1189 to conditionalize units based on the system architecture, as
1190 reported by uname()'s "machine" field.
1192 * systemd-networkd now supports matching on the system
1193 virtualization, architecture, kernel command line, host name
1196 * logind is now a lot more aggressive when suspending the
1197 machine due to a closed laptop lid. Instead of acting only
1198 on the lid close action, it will continuously watch the lid
1199 status and act on it. This is useful for laptops where the
1200 power button is on the outside of the chassis so that it can
1201 be reached without opening the lid (such as the Lenovo
1202 Yoga). On those machines, logind will now immediately
1203 re-suspend the machine if the power button has been
1204 accidentally pressed while the laptop was suspended and in a
1205 backpack or similar.
1207 * logind will now watch SW_DOCK switches and inhibit reaction
1208 to the lid switch if it is pressed. This means that logind
1209 will not suspend the machine anymore if the lid is closed
1210 and the system is docked, if the laptop supports SW_DOCK
1211 notifications via the input layer. Note that ACPI docking
1212 stations do not generate this currently. Also note that this
1213 logic is usually not fully sufficient and Desktop
1214 Environments should take a lid switch inhibitor lock when an
1215 external display is connected, as systemd will not watch
1218 * nspawn will now make use of the devices cgroup controller by
1219 default, and only permit creation of and access to the usual
1220 API device nodes like /dev/null or /dev/random, as well as
1221 access to (but not creation of) the pty devices.
1223 * We will now ship a default .network file for
1224 systemd-networkd that automatically configures DHCP for
1225 network interfaces created by nspawn's --network-veth or
1226 --network-bridge= switches.
1228 * systemd will now understand the usual M, K, G, T suffixes
1229 according to SI conventions (i.e. to the base 1000) when
1230 referring to throughput and hardware metrics. It will stay
1231 with IEC conventions (i.e. to the base 1024) for software
1232 metrics, according to what is customary according to
1233 Wikipedia. We explicitly document which base applies for
1234 each configuration option.
1236 * The DeviceAllow= setting in unit files now supports a syntax
1237 to whitelist an entire group of devices node majors at once,
1238 based on the /proc/devices listing. For example, with the
1239 string "char-pts", it is now possible to whitelist all
1240 current and future pseudo-TTYs at once.
1242 * sd-event learned a new "post" event source. Event sources of
1243 this type are triggered by the dispatching of any event
1244 source of a type that is not "post". This is useful for
1245 implementing clean-up and check event sources that are
1246 triggered by other work being done in the program.
1248 * systemd-networkd is no longer statically enabled, but uses
1249 the usual [Install] sections so that it can be
1250 enabled/disabled using systemctl. It still is enabled by
1253 * When creating a veth interface pair with systemd-nspawn, the
1254 host side will now be prefixed with "vb-" if
1255 --network-bridge= is used, and with "ve-" if --network-veth
1256 is used. This way, it is easy to distinguish these cases on
1257 the host, for example to apply different configuration to
1258 them with systemd-networkd.
1260 * The compatibility libraries for libsystemd-journal.so,
1261 libsystem-id128.so, libsystemd-login.so and
1262 libsystemd-daemon.so do not make use of IFUNC
1263 anymore. Instead, we now build libsystemd.so multiple times
1264 under these alternative names. This means that the footprint
1265 is drastically increased, but given that these are
1266 transitional compatibility libraries, this should not matter
1267 much. This change has been made necessary to support the ARM
1268 platform for these compatibility libraries, as the ARM
1269 toolchain is not really at the same level as the toolchain
1270 for other architectures like x86 and does not support
1271 IFUNC. Please make sure to use --enable-compat-libs only
1272 during a transitional period!
1274 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
1275 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1276 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
1277 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
1278 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
1279 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
1280 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
1281 Zbigniew Jędrzejewski-Szmek
1283 -- Berlin, 2014-02-24
1287 * A new component "systemd-networkd" has been added that can
1288 be used to configure local network interfaces statically or
1289 via DHCP. It is capable of bringing up bridges, VLANs, and
1290 bonding. Currently, no hook-ups for interactive network
1291 configuration are provided. Use this for your initrd,
1292 container, embedded, or server setup if you need a simple,
1293 yet powerful, network configuration solution. This
1294 configuration subsystem is quite nifty, as it allows wildcard
1295 hotplug matching in interfaces. For example, with a single
1296 configuration snippet, you can configure that all Ethernet
1297 interfaces showing up are automatically added to a bridge,
1298 or similar. It supports link-sensing and more.
1300 * A new tool "systemd-socket-proxyd" has been added which can
1301 act as a bidirectional proxy for TCP sockets. This is
1302 useful for adding socket activation support to services that
1303 do not actually support socket activation, including virtual
1304 machines and the like.
1306 * Add a new tool to save/restore rfkill state on
1309 * Save/restore state of keyboard backlights in addition to
1310 display backlights on shutdown/boot.
1312 * udev learned a new SECLABEL{} construct to label device
1313 nodes with a specific security label when they appear. For
1314 now, only SECLABEL{selinux} is supported, but the syntax is
1315 prepared for additional security frameworks.
1317 * udev gained a new scheme to configure link-level attributes
1318 from files in /etc/systemd/network/*.link. These files can
1319 match against MAC address, device path, driver name and type,
1320 and will apply attributes like the naming policy, link speed,
1321 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
1322 address assignment policy (randomized, ...).
1324 * The configuration of network interface naming rules for
1325 "permanent interface names" has changed: a new NamePolicy=
1326 setting in the [Link] section of .link files determines the
1327 priority of possible naming schemes (onboard, slot, mac,
1328 path). The default value of this setting is determined by
1329 /usr/lib/net/links/99-default.link. Old
1330 80-net-name-slot.rules udev configuration file has been
1331 removed, so local configuration overriding this file should
1332 be adapated to override 99-default.link instead.
1334 * When the User= switch is used in a unit file, also
1335 initialize $SHELL= based on the user database entry.
1337 * systemd no longer depends on libdbus. All communication is
1338 now done with sd-bus, systemd's low-level bus library
1341 * kdbus support has been added to PID 1 itself. When kdbus is
1342 enabled, this causes PID 1 to set up the system bus and
1343 enable support for a new ".busname" unit type that
1344 encapsulates bus name activation on kdbus. It works a little
1345 bit like ".socket" units, except for bus names. A new
1346 generator has been added that converts classic dbus1 service
1347 activation files automatically into native systemd .busname
1350 * sd-bus: add a light-weight vtable implementation that allows
1351 defining objects on the bus with a simple static const
1352 vtable array of its methods, signals and properties.
1354 * systemd will not generate or install static dbus
1355 introspection data anymore to /usr/share/dbus-1/interfaces,
1356 as the precise format of these files is unclear, and
1357 nothing makes use of it.
1359 * A proxy daemon is now provided to proxy clients connecting
1360 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1361 compatibility with classic D-Bus.
1363 * A bus driver implementation has been added that supports the
1364 classic D-Bus bus driver calls on kdbus, also for
1365 compatibility purposes.
1367 * A new API "sd-event.h" has been added that implements a
1368 minimal event loop API built around epoll. It provides a
1369 couple of features that direct epoll usage is lacking:
1370 prioritization of events, scales to large numbers of timer
1371 events, per-event timer slack (accuracy), system-wide
1372 coalescing of timer events, exit handlers, watchdog
1373 supervision support using systemd's sd_notify() API, child
1376 * A new API "sd-rntl.h" has been added that provides an API
1377 around the route netlink interface of the kernel, similar in
1378 style to "sd-bus.h".
1380 * A new API "sd-dhcp-client.h" has been added that provides a
1381 small DHCPv4 client-side implementation. This is used by
1384 * There is a new kernel command line option
1385 "systemd.restore_state=0|1". When set to "0", none of the
1386 systemd tools will restore saved runtime state to hardware
1387 devices. More specifically, the rfkill and backlight states
1390 * The FsckPassNo= compatibility option in mount/service units
1391 has been removed. The fstab generator will now add the
1392 necessary dependencies automatically, and does not require
1393 PID1's support for that anymore.
1395 * journalctl gained a new switch, --list-boots, that lists
1396 recent boots with their times and boot IDs.
1398 * The various tools like systemctl, loginctl, timedatectl,
1399 busctl, systemd-run, ... have gained a new switch "-M" to
1400 connect to a specific, local OS container (as direct
1401 connection, without requiring SSH). This works on any
1402 container that is registered with machined, such as those
1403 created by libvirt-lxc or nspawn.
1405 * systemd-run and systemd-analyze also gained support for "-H"
1406 to connect to remote hosts via SSH. This is particularly
1407 useful for systemd-run because it enables queuing of jobs
1408 onto remote systems.
1410 * machinectl gained a new command "login" to open a getty
1411 login in any local container. This works with any container
1412 that is registered with machined (such as those created by
1413 libvirt-lxc or nspawn), and which runs systemd inside.
1415 * machinectl gained a new "reboot" command that may be used to
1416 trigger a reboot on a specific container that is registered
1417 with machined. This works on any container that runs an init
1418 system of some kind.
1420 * systemctl gained a new "list-timers" command to print a nice
1421 listing of installed timer units with the times they elapse
1424 * Alternative reboot() parameters may now be specified on the
1425 "systemctl reboot" command line and are passed to the
1426 reboot() system call.
1428 * systemctl gained a new --job-mode= switch to configure the
1429 mode to queue a job with. This is a more generic version of
1430 --fail, --irreversible, and --ignore-dependencies, which are
1431 still available but not advertised anymore.
1433 * /etc/systemd/system.conf gained new settings to configure
1434 various default timeouts of units, as well as the default
1435 start limit interval and burst. These may still be overridden
1438 * PID1 will now export on the bus profile data of the security
1439 policy upload process (such as the SELinux policy upload to
1442 * journald: when forwarding logs to the console, include
1443 timestamps (following the setting in
1444 /sys/module/printk/parameters/time).
1446 * OnCalendar= in timer units now understands the special
1447 strings "yearly" and "annually". (Both are equivalent)
1449 * The accuracy of timer units is now configurable with the new
1450 AccuracySec= setting. It defaults to 1min.
1452 * A new dependency type JoinsNamespaceOf= has been added that
1453 allows running two services within the same /tmp and network
1454 namespace, if PrivateNetwork= or PrivateTmp= are used.
1456 * A new command "cat" has been added to systemctl. It outputs
1457 the original unit file of a unit, and concatenates the
1458 contents of additional "drop-in" unit file snippets, so that
1459 the full configuration is shown.
1461 * systemctl now supports globbing on the various "list-xyz"
1462 commands, like "list-units" or "list-sockets", as well as on
1463 those commands which take multiple unit names.
1465 * journalctl's --unit= switch gained support for globbing.
1467 * All systemd daemons now make use of the watchdog logic so
1468 that systemd automatically notices when they hang.
1470 * If the $container_ttys environment variable is set,
1471 getty-generator will automatically spawn a getty for each
1472 listed tty. This is useful for container managers to request
1473 login gettys to be spawned on as many ttys as needed.
1475 * %h, %s, %U specifier support is not available anymore when
1476 used in unit files for PID 1. This is because NSS calls are
1477 not safe from PID 1. They stay available for --user
1478 instances of systemd, and as special case for the root user.
1480 * loginctl gained a new "--no-legend" switch to turn off output
1483 * The "sd-login.h" API gained three new calls:
1484 sd_session_is_remote(), sd_session_get_remote_user(),
1485 sd_session_get_remote_host() to query information about
1488 * The udev hardware database now also carries vendor/product
1489 information of SDIO devices.
1491 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
1492 determine whether watchdog notifications are requested by
1495 * Socket-activated per-connection services now include a
1496 short description of the connection parameters in the
1499 * tmpfiles gained a new "--boot" option. When this is not used,
1500 only lines where the command character is not suffixed with
1501 "!" are executed. When this option is specified, those
1502 options are executed too. This partitions tmpfiles
1503 directives into those that can be safely executed at any
1504 time, and those which should be run only at boot (for
1505 example, a line that creates /run/nologin).
1507 * A new API "sd-resolve.h" has been added which provides a simple
1508 asynchronous wrapper around glibc NSS host name resolution
1509 calls, such as getaddrinfo(). In contrast to glibc's
1510 getaddrinfo_a(), it does not use signals. In contrast to most
1511 other asynchronous name resolution libraries, this one does
1512 not reimplement DNS, but reuses NSS, so that alternate
1513 host name resolution systems continue to work, such as mDNS,
1514 LDAP, etc. This API is based on libasyncns, but it has been
1515 cleaned up for inclusion in systemd.
1517 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
1518 "sd-daemon.h" are no longer found in individual libraries
1519 libsystemd-journal.so, libsystemd-login.so,
1520 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
1521 merged them into a single library, libsystemd.so, which
1522 provides all symbols. The reason for this is cyclic
1523 dependencies, as these libraries tend to use each other's
1524 symbols. So far, we have managed to workaround that by linking
1525 a copy of a good part of our code into each of these
1526 libraries again and again, which, however, makes certain
1527 things hard to do, like sharing static variables. Also, it
1528 substantially increases footprint. With this change, there
1529 is only one library for the basic APIs systemd
1530 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
1531 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
1532 library as well, however are subject to the --enable-kdbus
1533 switch (see below). Note that "sd-dhcp-client.h" is not part
1534 of this library (this is because it only consumes, never
1535 provides, services of/to other APIs). To make the transition
1536 easy from the separate libraries to the unified one, we
1537 provide the --enable-compat-libs compile-time switch which
1538 will generate stub libraries that are compatible with the
1539 old ones but redirect all calls to the new one.
1541 * All of the kdbus logic and the new APIs "sd-bus.h",
1542 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
1543 and "sd-utf8.h" are compile-time optional via the
1544 "--enable-kdbus" switch, and they are not compiled in by
1545 default. To make use of kdbus, you have to explicitly enable
1546 the switch. Note however, that neither the kernel nor the
1547 userspace API for all of this is considered stable yet. We
1548 want to maintain the freedom to still change the APIs for
1549 now. By specifying this build-time switch, you acknowledge
1550 that you are aware of the instability of the current
1553 * Also, note that while kdbus is pretty much complete,
1554 it lacks one thing: proper policy support. This means you
1555 can build a fully working system with all features; however,
1556 it will be highly insecure. Policy support will be added in
1557 one of the next releases, at the same time that we will
1558 declare the APIs stable.
1560 * When the kernel command-line argument "kdbus" is specified,
1561 systemd will automatically load the kdbus.ko kernel module. At
1562 this stage of development, it is only useful for testing kdbus
1563 and should not be used in production. Note: if "--enable-kdbus"
1564 is specified, and the kdbus.ko kernel module is available, and
1565 "kdbus" is added to the kernel command line, the entire system
1566 runs with kdbus instead of dbus-daemon, with the above mentioned
1567 problem of missing the system policy enforcement. Also a future
1568 version of kdbus.ko or a newer systemd will not be compatible with
1569 each other, and will unlikely be able to boot the machine if only
1570 one of them is updated.
1572 * systemctl gained a new "import-environment" command which
1573 uploads the caller's environment (or parts thereof) into the
1574 service manager so that it is inherited by services started
1575 by the manager. This is useful to upload variables like
1576 $DISPLAY into the user service manager.
1578 * A new PrivateDevices= switch has been added to service units
1579 which allows running a service with a namespaced /dev
1580 directory that does not contain any device nodes for
1581 physical devices. More specifically, it only includes devices
1582 such as /dev/null, /dev/urandom, and /dev/zero which are API
1585 * logind has been extended to support behaviour like VT
1586 switching on seats that do not support a VT. This makes
1587 multi-session available on seats that are not the first seat
1588 (seat0), and on systems where kernel support for VTs has
1589 been disabled at compile-time.
1591 * If a process holds a delay lock for system sleep or shutdown
1592 and fails to release it in time, we will now log its
1593 identity. This makes it easier to identify processes that
1594 cause slow suspends or power-offs.
1596 * When parsing /etc/crypttab, support for a new key-slot=
1597 option as supported by Debian is added. It allows indicating
1598 which LUKS slot to use on disk, speeding up key loading.
1600 * The sd_journald_sendv() API call has been checked and
1601 officially declared to be async-signal-safe so that it may
1602 be invoked from signal handlers for logging purposes.
1604 * Boot-time status output is now enabled automatically after a
1605 short timeout if boot does not progress, in order to give
1606 the user an indication what she or he is waiting for.
1608 * The boot-time output has been improved to show how much time
1609 remains until jobs expire.
1611 * The KillMode= switch in service units gained a new possible
1612 value "mixed". If set, and the unit is shut down, then the
1613 initial SIGTERM signal is sent only to the main daemon
1614 process, while the following SIGKILL signal is sent to
1615 all remaining processes of the service.
1617 * When a scope unit is registered, a new property "Controller"
1618 may be set. If set to a valid bus name, systemd will send a
1619 RequestStop() signal to this name when it would like to shut
1620 down the scope. This may be used to hook manager logic into
1621 the shutdown logic of scope units. Also, scope units may now
1622 be put in a special "abandoned" state, in which case the
1623 manager process which created them takes no further
1624 responsibilities for it.
1626 * When reading unit files, systemd will now verify
1627 the access mode of these files, and warn about certain
1628 suspicious combinations. This has been added to make it
1629 easier to track down packaging bugs where unit files are
1630 marked executable or world-writable.
1632 * systemd-nspawn gained a new "--setenv=" switch to set
1633 container-wide environment variables. The similar option in
1634 systemd-activate was renamed from "--environment=" to
1635 "--setenv=" for consistency.
1637 * systemd-nspawn has been updated to create a new kdbus domain
1638 for each container that is invoked, thus allowing each
1639 container to have its own set of system and user buses,
1640 independent of the host.
1642 * systemd-nspawn gained a new --drop-capability= switch to run
1643 the container with less capabilities than the default. Both
1644 --drop-capability= and --capability= now take the special
1645 string "all" for dropping or keeping all capabilities.
1647 * systemd-nspawn gained new switches for executing containers
1648 with specific SELinux labels set.
1650 * systemd-nspawn gained a new --quiet switch to not generate
1651 any additional output but the container's own console
1654 * systemd-nspawn gained a new --share-system switch to run a
1655 container without PID namespacing enabled.
1657 * systemd-nspawn gained a new --register= switch to control
1658 whether the container is registered with systemd-machined or
1659 not. This is useful for containers that do not run full
1660 OS images, but only specific apps.
1662 * systemd-nspawn gained a new --keep-unit which may be used
1663 when invoked as the only program from a service unit, and
1664 results in registration of the unit service itself in
1665 systemd-machined, instead of a newly opened scope unit.
1667 * systemd-nspawn gained a new --network-interface= switch for
1668 moving arbitrary interfaces to the container. The new
1669 --network-veth switch creates a virtual Ethernet connection
1670 between host and container. The new --network-bridge=
1671 switch then allows assigning the host side of this virtual
1672 Ethernet connection to a bridge device.
1674 * systemd-nspawn gained a new --personality= switch for
1675 setting the kernel personality for the container. This is
1676 useful when running a 32-bit container on a 64-bit host. A
1677 similar option Personality= is now also available for service
1680 * logind will now also track a "Desktop" identifier for each
1681 session which encodes the desktop environment of it. This is
1682 useful for desktop environments that want to identify
1683 multiple running sessions of itself easily.
1685 * A new SELinuxContext= setting for service units has been
1686 added that allows setting a specific SELinux execution
1687 context for a service.
1689 * Most systemd client tools will now honour $SYSTEMD_LESS for
1690 settings of the "less" pager. By default, these tools will
1691 override $LESS to allow certain operations to work, such as
1692 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1693 influence this logic.
1695 * systemd's "seccomp" hook-up has been changed to make use of
1696 the libseccomp library instead of using its own
1697 implementation. This has benefits for portability among
1700 * For usage together with SystemCallFilter=, a new
1701 SystemCallErrorNumber= setting has been introduced that
1702 allows configuration of a system error number to be returned
1703 on filtered system calls, instead of immediately killing the
1704 process. Also, SystemCallArchitectures= has been added to
1705 limit access to system calls of a particular architecture
1706 (in order to turn off support for unused secondary
1707 architectures). There is also a global
1708 SystemCallArchitectures= setting in system.conf now to turn
1709 off support for non-native system calls system-wide.
1711 * systemd requires a kernel with a working name_to_handle_at(),
1712 please see the kernel config requirements in the README file.
1714 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1715 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1716 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1717 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1718 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1719 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1720 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1721 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1722 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1723 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1724 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1725 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1726 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1727 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1728 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1729 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1730 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1731 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1732 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1733 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1734 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1735 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1736 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1737 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1739 -- Berlin, 2014-02-20
1743 * logind has gained support for facilitating privileged input
1744 and drm device access for unprivileged clients. This work is
1745 useful to allow Wayland display servers (and similar
1746 programs, such as kmscon) to run under the user's ID and
1747 access input and drm devices which are normally
1748 protected. When this is used (and the kernel is new enough)
1749 logind will "mute" IO on the file descriptors passed to
1750 Wayland as long as it is in the background and "unmute" it
1751 if it returns into the foreground. This allows secure
1752 session switching without allowing background sessions to
1753 eavesdrop on input and display data. This also introduces
1754 session switching support if VT support is turned off in the
1755 kernel, and on seats that are not seat0.
1757 * A new kernel command line option luks.options= is understood
1758 now which allows specifying LUKS options for usage for LUKS
1759 encrypted partitions specified with luks.uuid=.
1761 * tmpfiles.d(5) snippets may now use specifier expansion in
1762 path names. More specifically %m, %b, %H, %v, are now
1763 replaced by the local machine id, boot id, hostname, and
1764 kernel version number.
1766 * A new tmpfiles.d(5) command "m" has been introduced which
1767 may be used to change the owner/group/access mode of a file
1768 or directory if it exists, but do nothing if it does not.
1770 * This release removes high-level support for the
1771 MemorySoftLimit= cgroup setting. The underlying kernel
1772 cgroup attribute memory.soft_limit= is currently badly
1773 designed and likely to be removed from the kernel API in its
1774 current form, hence we should not expose it for now.
1776 * The memory.use_hierarchy cgroup attribute is now enabled for
1777 all cgroups systemd creates in the memory cgroup
1778 hierarchy. This option is likely to be come the built-in
1779 default in the kernel anyway, and the non-hierarchial mode
1780 never made much sense in the intrinsically hierarchial
1783 * A new field _SYSTEMD_SLICE= is logged along with all journal
1784 messages containing the slice a message was generated
1785 from. This is useful to allow easy per-customer filtering of
1786 logs among other things.
1788 * systemd-journald will no longer adjust the group of journal
1789 files it creates to the "systemd-journal" group. Instead we
1790 rely on the journal directory to be owned by the
1791 "systemd-journal" group, and its setgid bit set, so that the
1792 kernel file system layer will automatically enforce that
1793 journal files inherit this group assignment. The reason for
1794 this change is that we cannot allow NSS look-ups from
1795 journald which would be necessary to resolve
1796 "systemd-journal" to a numeric GID, because this might
1797 create deadlocks if NSS involves synchronous queries to
1798 other daemons (such as nscd, or sssd) which in turn are
1799 logging clients of journald and might block on it, which
1800 would then dead lock. A tmpfiles.d(5) snippet included in
1801 systemd will make sure the setgid bit and group are
1802 properly set on the journal directory if it exists on every
1803 boot. However, we recommend adjusting it manually after
1804 upgrades too (or from RPM scriptlets), so that the change is
1805 not delayed until next reboot.
1807 * Backlight and random seed files in /var/lib/ have moved into
1808 the /var/lib/systemd/ directory, in order to centralize all
1809 systemd generated files in one directory.
1811 * Boot time performance measurements (as displayed by
1812 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1813 performance information if that's available to determine how
1814 much time BIOS and boot loader initialization required. With
1815 a sufficiently new BIOS you hence no longer need to boot
1816 with Gummiboot to get access to such information.
1818 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1819 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1820 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1821 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1822 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1823 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1824 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1826 -- Berlin, 2013-10-02
1830 * The Restart= option for services now understands a new
1831 on-watchdog setting, which will restart the service
1832 automatically if the service stops sending out watchdog keep
1833 alive messages (as configured with WatchdogSec=).
1835 * The getty generator (which is responsible for bringing up a
1836 getty on configured serial consoles) will no longer only
1837 start a getty on the primary kernel console but on all
1838 others, too. This makes the order in which console= is
1839 specified on the kernel command line less important.
1841 * libsystemd-logind gained a new sd_session_get_vt() call to
1842 retrieve the VT number of a session.
1844 * If the option "tries=0" is set for an entry of /etc/crypttab
1845 its passphrase is queried indefinitely instead of any
1846 maximum number of tries.
1848 * If a service with a configure PID file terminates its PID
1849 file will now be removed automatically if it still exists
1850 afterwards. This should put an end to stale PID files.
1852 * systemd-run will now also take relative binary path names
1853 for execution and no longer insists on absolute paths.
1855 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1856 paths that are optionally prefixed with "-" to indicate that
1857 it should not be considered a failure if they do not exist.
1859 * journalctl -o (and similar commands) now understands a new
1860 output mode "short-precise", it is similar to "short" but
1861 shows timestamps with usec accuracy.
1863 * The option "discard" (as known from Debian) is now
1864 synonymous to "allow-discards" in /etc/crypttab. In fact,
1865 "discard" is preferred now (since it is easier to remember
1868 * Some licensing clean-ups were made, so that more code is now
1869 LGPL-2.1 licensed than before.
1871 * A minimal tool to save/restore the display backlight
1872 brightness across reboots has been added. It will store the
1873 backlight setting as late as possible at shutdown, and
1874 restore it as early as possible during reboot.
1876 * A logic to automatically discover and enable home and swap
1877 partitions on GPT disks has been added. With this in place
1878 /etc/fstab becomes optional for many setups as systemd can
1879 discover certain partitions located on the root disk
1880 automatically. Home partitions are recognized under their
1881 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1882 partitions are recognized under their GPT type ID
1883 0657fd6da4ab43c484e50933c84b4f4f.
1885 * systemd will no longer pass any environment from the kernel
1886 or initrd to system services. If you want to set an
1887 environment for all services, do so via the kernel command
1888 line systemd.setenv= assignment.
1890 * The systemd-sysctl tool no longer natively reads the file
1891 /etc/sysctl.conf. If desired, the file should be symlinked
1892 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1893 legacy support by a symlink rather than built-in code, it
1894 also makes the otherwise hidden order of application of the
1895 different files visible. (Note that this partly reverts to a
1896 pre-198 application order of sysctl knobs!)
1898 * The "systemctl set-log-level" and "systemctl dump" commands
1899 have been moved to systemd-analyze.
1901 * systemd-run learned the new --remain-after-exit switch,
1902 which causes the scope unit not to be cleaned up
1903 automatically after the process terminated.
1905 * tmpfiles learned a new --exclude-prefix= switch to exclude
1906 certain paths from operation.
1908 * journald will now automatically flush all messages to disk
1909 as soon as a message of the log priorities CRIT, ALERT or
1912 Contributions from: Andrew Cook, Brandon Philips, Christian
1913 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1914 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1915 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1916 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1917 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1918 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1919 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1920 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1921 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1922 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1923 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1924 William Giokas, Zbigniew Jędrzejewski-Szmek
1926 -- Berlin, 2013-09-13
1930 * The documentation has been updated to cover the various new
1931 concepts introduced with 205.
1933 * Unit files now understand the new %v specifier which
1934 resolves to the kernel version string as returned by "uname
1937 * systemctl now supports filtering the unit list output by
1938 load state, active state and sub state, using the new
1941 * "systemctl status" will now show the results of the
1942 condition checks (like ConditionPathExists= and similar) of
1943 the last start attempts of the unit. They are also logged to
1946 * "journalctl -b" may now be used to look for boot output of a
1947 specific boot. Try "journalctl -b -1" for the previous boot,
1948 but the syntax is substantially more powerful.
1950 * "journalctl --show-cursor" has been added which prints the
1951 cursor string the last shown log line. This may then be used
1952 with the new "journalctl --after-cursor=" switch to continue
1953 browsing logs from that point on.
1955 * "journalctl --force" may now be used to force regeneration
1958 * Creation of "dead" device nodes has been moved from udev
1959 into kmod and tmpfiles. Previously, udev would read the kmod
1960 databases to pre-generate dead device nodes based on meta
1961 information contained in kernel modules, so that these would
1962 be auto-loaded on access rather then at boot. As this
1963 does not really have much to do with the exposing actual
1964 kernel devices to userspace this has always been slightly
1965 alien in the udev codebase. Following the new scheme kmod
1966 will now generate a runtime snippet for tmpfiles from the
1967 module meta information and it now is tmpfiles' job to the
1968 create the nodes. This also allows overriding access and
1969 other parameters for the nodes using the usual tmpfiles
1970 facilities. As side effect this allows us to remove the
1971 CAP_SYS_MKNOD capability bit from udevd entirely.
1973 * logind's device ACLs may now be applied to these "dead"
1974 devices nodes too, thus finally allowing managed access to
1975 devices such as /dev/snd/sequencer whithout loading the
1976 backing module right-away.
1978 * A new RPM macro has been added that may be used to apply
1979 tmpfiles configuration during package installation.
1981 * systemd-detect-virt and ConditionVirtualization= now can
1982 detect User-Mode-Linux machines (UML).
1984 * journald will now implicitly log the effective capabilities
1985 set of processes in the message metadata.
1987 * systemd-cryptsetup has gained support for TrueCrypt volumes.
1989 * The initrd interface has been simplified (more specifically,
1990 support for passing performance data via environment
1991 variables and fsck results via files in /run has been
1992 removed). These features were non-essential, and are
1993 nowadays available in a much nicer way by having systemd in
1994 the initrd serialize its state and have the hosts systemd
1995 deserialize it again.
1997 * The udev "keymap" data files and tools to apply keyboard
1998 specific mappings of scan to key codes, and force-release
1999 scan code lists have been entirely replaced by a udev
2000 "keyboard" builtin and a hwdb data file.
2002 * systemd will now honour the kernel's "quiet" command line
2003 argument also during late shutdown, resulting in a
2004 completely silent shutdown when used.
2006 * There's now an option to control the SO_REUSEPORT socket
2007 option in .socket units.
2009 * Instance units will now automatically get a per-template
2010 subslice of system.slice unless something else is explicitly
2011 configured. For example, instances of sshd@.service will now
2012 implicitly be placed in system-sshd.slice rather than
2013 system.slice as before.
2015 * Test coverage support may now be enabled at build time.
2017 Contributions from: Dave Reisner, Frederic Crozat, Harald
2018 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
2019 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
2020 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
2021 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
2022 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
2023 Giokas, Zbigniew Jędrzejewski-Szmek
2025 -- Berlin, 2013-07-23
2029 * Two new unit types have been introduced:
2031 Scope units are very similar to service units, however, are
2032 created out of pre-existing processes -- instead of PID 1
2033 forking off the processes. By using scope units it is
2034 possible for system services and applications to group their
2035 own child processes (worker processes) in a powerful way
2036 which then maybe used to organize them, or kill them
2037 together, or apply resource limits on them.
2039 Slice units may be used to partition system resources in an
2040 hierarchial fashion and then assign other units to them. By
2041 default there are now three slices: system.slice (for all
2042 system services), user.slice (for all user sessions),
2043 machine.slice (for VMs and containers).
2045 Slices and scopes have been introduced primarily in
2046 context of the work to move cgroup handling to a
2047 single-writer scheme, where only PID 1
2048 creates/removes/manages cgroups.
2050 * There's a new concept of "transient" units. In contrast to
2051 normal units these units are created via an API at runtime,
2052 not from configuration from disk. More specifically this
2053 means it is now possible to run arbitrary programs as
2054 independent services, with all execution parameters passed
2055 in via bus APIs rather than read from disk. Transient units
2056 make systemd substantially more dynamic then it ever was,
2057 and useful as a general batch manager.
2059 * logind has been updated to make use of scope and slice units
2060 for managing user sessions. As a user logs in he will get
2061 his own private slice unit, to which all sessions are added
2062 as scope units. We also added support for automatically
2063 adding an instance of user@.service for the user into the
2064 slice. Effectively logind will no longer create cgroup
2065 hierarchies on its own now, it will defer entirely to PID 1
2066 for this by means of scope, service and slice units. Since
2067 user sessions this way become entities managed by PID 1
2068 the output of "systemctl" is now a lot more comprehensive.
2070 * A new mini-daemon "systemd-machined" has been added which
2071 may be used by virtualization managers to register local
2072 VMs/containers. nspawn has been updated accordingly, and
2073 libvirt will be updated shortly. machined will collect a bit
2074 of meta information about the VMs/containers, and assign
2075 them their own scope unit (see above). The collected
2076 meta-data is then made available via the "machinectl" tool,
2077 and exposed in "ps" and similar tools. machined/machinectl
2078 is compile-time optional.
2080 * As discussed earlier, the low-level cgroup configuration
2081 options ControlGroup=, ControlGroupModify=,
2082 ControlGroupPersistent=, ControlGroupAttribute= have been
2083 removed. Please use high-level attribute settings instead as
2084 well as slice units.
2086 * A new bus call SetUnitProperties() has been added to alter
2087 various runtime parameters of a unit. This is primarily
2088 useful to alter cgroup parameters dynamically in a nice way,
2089 but will be extended later on to make more properties
2090 modifiable at runtime. systemctl gained a new set-properties
2091 command that wraps this call.
2093 * A new tool "systemd-run" has been added which can be used to
2094 run arbitrary command lines as transient services or scopes,
2095 while configuring a number of settings via the command
2096 line. This tool is currently very basic, however already
2097 very useful. We plan to extend this tool to even allow
2098 queuing of execution jobs with time triggers from the
2099 command line, similar in fashion to "at".
2101 * nspawn will now inform the user explicitly that kernels with
2102 audit enabled break containers, and suggest the user to turn
2105 * Support for detecting the IMA and AppArmor security
2106 frameworks with ConditionSecurity= has been added.
2108 * journalctl gained a new "-k" switch for showing only kernel
2109 messages, mimicking dmesg output; in addition to "--user"
2110 and "--system" switches for showing only user's own logs
2113 * systemd-delta can now show information about drop-in
2114 snippets extending unit files.
2116 * libsystemd-bus has been substantially updated but is still
2117 not available as public API.
2119 * systemd will now look for the "debug" argument on the kernel
2120 command line and enable debug logging, similar to what
2121 "systemd.log_level=debug" already did before.
2123 * "systemctl set-default", "systemctl get-default" has been
2124 added to configure the default.target symlink, which
2125 controls what to boot into by default.
2127 * "systemctl set-log-level" has been added as a convenient
2128 way to raise and lower systemd logging threshold.
2130 * "systemd-analyze plot" will now show the time the various
2131 generators needed for execution, as well as information
2132 about the unit file loading.
2134 * libsystemd-journal gained a new sd_journal_open_files() call
2135 for opening specific journal files. journactl also gained a
2136 new switch to expose this new functionality. Previously we
2137 only supported opening all files from a directory, or all
2138 files from the system, as opening individual files only is
2139 racy due to journal file rotation.
2141 * systemd gained the new DefaultEnvironment= setting in
2142 /etc/systemd/system.conf to set environment variables for
2145 * If a privileged process logs a journal message with the
2146 OBJECT_PID= field set, then journald will automatically
2147 augment this with additional OBJECT_UID=, OBJECT_GID=,
2148 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2149 system services want to log events about specific client
2150 processes. journactl/systemctl has been updated to make use
2151 of this information if all log messages regarding a specific
2154 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2155 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2156 Reisner, David Coppa, David King, David Strauss, Eelco
2157 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2158 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2159 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2160 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2161 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2162 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2163 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2164 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2165 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2166 Łukasz Stelmach, 장동준
2170 * The Python bindings gained some minimal support for the APIs
2171 exposed by libsystemd-logind.
2173 * ConditionSecurity= gained support for detecting SMACK. Since
2174 this condition already supports SELinux and AppArmor we only
2175 miss IMA for this. Patches welcome!
2177 Contributions from: Karol Lewandowski, Lennart Poettering,
2178 Zbigniew Jędrzejewski-Szmek
2182 * systemd-nspawn will now create /etc/resolv.conf if
2183 necessary, before bind-mounting the host's file onto it.
2185 * systemd-nspawn will now store meta information about a
2186 container on the container's cgroup as extended attribute
2187 fields, including the root directory.
2189 * The cgroup hierarchy has been reworked in many ways. All
2190 objects any of the components systemd creates in the cgroup
2191 tree are now suffixed. More specifically, user sessions are
2192 now placed in cgroups suffixed with ".session", users in
2193 cgroups suffixed with ".user", and nspawn containers in
2194 cgroups suffixed with ".nspawn". Furthermore, all cgroup
2195 names are now escaped in a simple scheme to avoid collision
2196 of userspace object names with kernel filenames. This work
2197 is preparation for making these objects relocatable in the
2198 cgroup tree, in order to allow easy resource partitioning of
2199 these objects without causing naming conflicts.
2201 * systemctl list-dependencies gained the new switches
2202 --plain, --reverse, --after and --before.
2204 * systemd-inhibit now shows the process name of processes that
2205 have taken an inhibitor lock.
2207 * nss-myhostname will now also resolve "localhost"
2208 implicitly. This makes /etc/hosts an optional file and
2209 nicely handles that on IPv6 ::1 maps to both "localhost" and
2212 * libsystemd-logind.so gained a new call
2213 sd_get_machine_names() to enumerate running containers and
2214 VMs (currently only supported by very new libvirt and
2215 nspawn). sd_login_monitor can now be used to watch
2216 VMs/containers coming and going.
2218 * .include is not allowed recursively anymore, and only in
2219 unit files. Usually it is better to use drop-in snippets in
2220 .d/*.conf anyway, as introduced with systemd 198.
2222 * systemd-analyze gained a new "critical-chain" command that
2223 determines the slowest chain of units run during system
2224 boot-up. It is very useful for tracking down where
2225 optimizing boot time is the most beneficial.
2227 * systemd will no longer allow manipulating service paths in
2228 the name=systemd:/system cgroup tree using ControlGroup= in
2229 units. (But is still fine with it in all other dirs.)
2231 * There's a new systemd-nspawn@.service service file that may
2232 be used to easily run nspawn containers as system
2233 services. With the container's root directory in
2234 /var/lib/container/foobar it is now sufficient to run
2235 "systemctl start systemd-nspawn@foobar.service" to boot it.
2237 * systemd-cgls gained a new parameter "--machine" to list only
2238 the processes within a certain container.
2240 * ConditionSecurity= now can check for "apparmor". We still
2241 are lacking checks for SMACK and IMA for this condition
2242 check though. Patches welcome!
2244 * A new configuration file /etc/systemd/sleep.conf has been
2245 added that may be used to configure which kernel operation
2246 systemd is supposed to execute when "suspend", "hibernate"
2247 or "hybrid-sleep" is requested. This makes the new kernel
2248 "freeze" state accessible to the user.
2250 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
2251 the passed argument if applicable.
2253 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
2254 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
2255 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
2256 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
2257 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
2258 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
2259 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
2264 * The output of 'systemctl list-jobs' got some polishing. The
2265 '--type=' argument may now be passed more than once. A new
2266 command 'systemctl list-sockets' has been added which shows
2267 a list of kernel sockets systemd is listening on with the
2268 socket units they belong to, plus the units these socket
2271 * The experimental libsystemd-bus library got substantial
2272 updates to work in conjunction with the (also experimental)
2273 kdbus kernel project. It works well enough to exchange
2274 messages with some sophistication. Note that kdbus is not
2275 ready yet, and the library is mostly an elaborate test case
2276 for now, and not installable.
2278 * systemd gained a new unit 'systemd-static-nodes.service'
2279 that generates static device nodes earlier during boot, and
2280 can run in conjunction with udev.
2282 * libsystemd-login gained a new call sd_pid_get_user_unit()
2283 to retrieve the user systemd unit a process is running
2284 in. This is useful for systems where systemd is used as
2287 * systemd-nspawn now places all containers in the new /machine
2288 top-level cgroup directory in the name=systemd
2289 hierarchy. libvirt will soon do the same, so that we get a
2290 uniform separation of /system, /user and /machine for system
2291 services, user processes and containers/virtual
2292 machines. This new cgroup hierarchy is also useful to stick
2293 stable names to specific container instances, which can be
2294 recognized later this way (this name may be controlled
2295 via systemd-nspawn's new -M switch). libsystemd-login also
2296 gained a new call sd_pid_get_machine_name() to retrieve the
2297 name of the container/VM a specific process belongs to.
2299 * bootchart can now store its data in the journal.
2301 * libsystemd-journal gained a new call
2302 sd_journal_add_conjunction() for AND expressions to the
2303 matching logic. This can be used to express more complex
2304 logical expressions.
2306 * journactl can now take multiple --unit= and --user-unit=
2309 * The cryptsetup logic now understands the "luks.key=" kernel
2310 command line switch for specifying a file to read the
2311 decryption key from. Also, if a configured key file is not
2312 found the tool will now automatically fall back to prompting
2315 * Python systemd.journal module was updated to wrap recently
2316 added functions from libsystemd-journal. The interface was
2317 changed to bring the low level interface in s.j._Reader
2318 closer to the C API, and the high level interface in
2319 s.j.Reader was updated to wrap and convert all data about
2322 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
2323 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
2324 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
2325 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
2326 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
2327 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2331 * journalctl --update-catalog now understands a new --root=
2332 option to operate on catalogs found in a different root
2335 * During shutdown after systemd has terminated all running
2336 services a final killing loop kills all remaining left-over
2337 processes. We will now print the name of these processes
2338 when we send SIGKILL to them, since this usually indicates a
2341 * If /etc/crypttab refers to password files stored on
2342 configured mount points automatic dependencies will now be
2343 generated to ensure the specific mount is established first
2344 before the key file is attempted to be read.
2346 * 'systemctl status' will now show information about the
2347 network sockets a socket unit is listening on.
2349 * 'systemctl status' will also shown information about any
2350 drop-in configuration file for units. (Drop-In configuration
2351 files in this context are files such as
2352 /etc/systemd/systemd/foobar.service.d/*.conf)
2354 * systemd-cgtop now optionally shows summed up CPU times of
2355 cgroups. Press '%' while running cgtop to switch between
2356 percentage and absolute mode. This is useful to determine
2357 which cgroups use up the most CPU time over the entire
2358 runtime of the system. systemd-cgtop has also been updated
2359 to be 'pipeable' for processing with further shell tools.
2361 * 'hostnamectl set-hostname' will now allow setting of FQDN
2364 * The formatting and parsing of time span values has been
2365 changed. The parser now understands fractional expressions
2366 such as "5.5h". The formatter will now output fractional
2367 expressions for all time spans under 1min, i.e. "5.123456s"
2368 rather than "5s 123ms 456us". For time spans under 1s
2369 millisecond values are shown, for those under 1ms
2370 microsecond values are shown. This should greatly improve
2371 all time-related output of systemd.
2373 * libsystemd-login and libsystemd-journal gained new
2374 functions for querying the poll() events mask and poll()
2375 timeout value for integration into arbitrary event
2378 * localectl gained the ability to list available X11 keymaps
2379 (models, layouts, variants, options).
2381 * 'systemd-analyze dot' gained the ability to filter for
2382 specific units via shell-style globs, to create smaller,
2383 more useful graphs. I.e. it is now possible to create simple
2384 graphs of all the dependencies between only target units, or
2385 of all units that Avahi has dependencies with.
2387 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2388 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2389 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2390 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2391 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2392 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2393 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2397 * The boot-time readahead implementation for rotating media
2398 will now read the read-ahead data in multiple passes which
2399 consist of all read requests made in equidistant time
2400 intervals. This means instead of strictly reading read-ahead
2401 data in its physical order on disk we now try to find a
2402 middle ground between physical and access time order.
2404 * /etc/os-release files gained a new BUILD_ID= field for usage
2405 on operating systems that provide continuous builds of OS
2408 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
2409 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
2410 William Douglas, Zbigniew Jędrzejewski-Szmek
2414 * systemd-python gained an API exposing libsystemd-daemon.
2416 * The SMACK setup logic gained support for uploading CIPSO
2419 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
2420 ReadOnlyDirectories= and InaccessibleDirectories= has
2421 changed. The private /tmp and /var/tmp directories are now
2422 shared by all processes of a service (which means
2423 ExecStartPre= may now leave data in /tmp that ExecStart= of
2424 the same service can still access). When a service is
2425 stopped its temporary directories are immediately deleted
2426 (normal clean-up with tmpfiles is still done in addition to
2429 * By default, systemd will now set a couple of sysctl
2430 variables in the kernel: the safe sysrq options are turned
2431 on, IP route verification is turned on, and source routing
2432 disabled. The recently added hardlink and softlink
2433 protection of the kernel is turned on. These settings should
2434 be reasonably safe, and good defaults for all new systems.
2436 * The predictable network naming logic may now be turned off
2437 with a new kernel command line switch: net.ifnames=0.
2439 * A new libsystemd-bus module has been added that implements a
2440 pretty complete D-Bus client library. For details see:
2442 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
2444 * journald will now explicitly flush the journal files to disk
2445 at the latest 5min after each write. The file will then also
2446 be marked offline until the next write. This should increase
2447 reliability in case of a crash. The synchronization delay
2448 can be configured via SyncIntervalSec= in journald.conf.
2450 * There's a new remote-fs-setup.target unit that can be used
2451 to pull in specific services when at least one remote file
2452 system is to be mounted.
2454 * There are new targets timers.target and paths.target as
2455 canonical targets to pull user timer and path units in
2456 from. This complements sockets.target with a similar
2457 purpose for socket units.
2459 * libudev gained a new call udev_device_set_attribute_value()
2460 to set sysfs attributes of a device.
2462 * The udev daemon now sets the default number of worker
2463 processes executed in parallel based on the number of available
2464 CPUs instead of the amount of available RAM. This is supposed
2465 to provide a more reliable default and limit a too aggressive
2466 paralellism for setups with 1000s of devices connected.
2468 Contributions from: Auke Kok, Colin Walters, Cristian
2469 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
2470 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
2471 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
2472 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
2473 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
2474 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
2475 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
2476 Zbigniew Jędrzejewski-Szmek
2480 * Configuration of unit files may now be extended via drop-in
2481 files without having to edit/override the unit files
2482 themselves. More specifically, if the administrator wants to
2483 change one value for a service file foobar.service he can
2484 now do so by dropping in a configuration snippet into
2485 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
2486 will load all these snippets and apply them on top of the
2487 main unit configuration file, possibly extending or
2488 overriding its settings. Using these drop-in snippets is
2489 generally nicer than the two earlier options for changing
2490 unit files locally: copying the files from
2491 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
2492 them there; or creating a new file in /etc/systemd/system/
2493 that incorporates the original one via ".include". Drop-in
2494 snippets into these .d/ directories can be placed in any
2495 directory systemd looks for units in, and the usual
2496 overriding semantics between /usr/lib, /etc and /run apply
2499 * Most unit file settings which take lists of items can now be
2500 reset by assigning the empty string to them. For example,
2501 normally, settings such as Environment=FOO=BAR append a new
2502 environment variable assignment to the environment block,
2503 each time they are used. By assigning Environment= the empty
2504 string the environment block can be reset to empty. This is
2505 particularly useful with the .d/*.conf drop-in snippets
2506 mentioned above, since this adds the ability to reset list
2507 settings from vendor unit files via these drop-ins.
2509 * systemctl gained a new "list-dependencies" command for
2510 listing the dependencies of a unit recursively.
2512 * Inhibitors are now honored and listed by "systemctl
2513 suspend", "systemctl poweroff" (and similar) too, not only
2514 GNOME. These commands will also list active sessions by
2517 * Resource limits (as exposed by the various control group
2518 controllers) can now be controlled dynamically at runtime
2519 for all units. More specifically, you can now use a command
2520 like "systemctl set-cgroup-attr foobar.service cpu.shares
2521 2000" to alter the CPU shares a specific service gets. These
2522 settings are stored persistently on disk, and thus allow the
2523 administrator to easily adjust the resource usage of
2524 services with a few simple commands. This dynamic resource
2525 management logic is also available to other programs via the
2526 bus. Almost any kernel cgroup attribute and controller is
2529 * systemd-vconsole-setup will now copy all font settings to
2530 all allocated VTs, where it previously applied them only to
2533 * libsystemd-login gained the new sd_session_get_tty() API
2536 * This release drops support for a few legacy or
2537 distribution-specific LSB facility names when parsing init
2538 scripts: $x-display-manager, $mail-transfer-agent,
2539 $mail-transport-agent, $mail-transfer-agent, $smtp,
2540 $null. Also, the mail-transfer-agent.target unit backing
2541 this has been removed. Distributions which want to retain
2542 compatibility with this should carry the burden for
2543 supporting this themselves and patch support for these back
2544 in, if they really need to. Also, the facilities $syslog and
2545 $local_fs are now ignored, since systemd does not support
2546 early-boot LSB init scripts anymore, and these facilities
2547 are implied anyway for normal services. syslog.target has
2550 * There are new bus calls on PID1's Manager object for
2551 cancelling jobs, and removing snapshot units. Previously,
2552 both calls were only available on the Job and Snapshot
2555 * systemd-journal-gatewayd gained SSL support.
2557 * The various "environment" files, such as /etc/locale.conf
2558 now support continuation lines with a backslash ("\") as
2559 last character in the line, similarly in style (but different)
2560 to how this is supported in shells.
2562 * For normal user processes the _SYSTEMD_USER_UNIT= field is
2563 now implicitly appended to every log entry logged. systemctl
2564 has been updated to filter by this field when operating on a
2565 user systemd instance.
2567 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
2568 CAP_AUDIT_CONTROL capabilities to the capabilities set for
2569 the container. This makes it easier to boot unmodified
2570 Fedora systems in a container, which however still requires
2571 audit=0 to be passed on the kernel command line. Auditing in
2572 kernel and userspace is unfortunately still too broken in
2573 context of containers, hence we recommend compiling it out
2574 of the kernel or using audit=0. Hopefully this will be fixed
2575 one day for good in the kernel.
2577 * nspawn gained the new --bind= and --bind-ro= parameters to
2578 bind mount specific directories from the host into the
2581 * nspawn will now mount its own devpts file system instance
2582 into the container, in order not to leak pty devices from
2583 the host into the container.
2585 * systemd will now read the firmware boot time performance
2586 information from the EFI variables, if the used boot loader
2587 supports this, and takes it into account for boot performance
2588 analysis via "systemd-analyze". This is currently supported
2589 only in conjunction with Gummiboot, but could be supported
2590 by other boot loaders too. For details see:
2592 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
2594 * A new generator has been added that automatically mounts the
2595 EFI System Partition (ESP) to /boot, if that directory
2596 exists, is empty, and no other file system has been
2597 configured to be mounted there.
2599 * logind will now send out PrepareForSleep(false) out
2600 unconditionally, after coming back from suspend. This may be
2601 used by applications as asynchronous notification for
2602 system resume events.
2604 * "systemctl unlock-sessions" has been added, that allows
2605 unlocking the screens of all user sessions at once, similar
2606 to how "systemctl lock-sessions" already locked all users
2607 sessions. This is backed by a new D-Bus call UnlockSessions().
2609 * "loginctl seat-status" will now show the master device of a
2610 seat. (i.e. the device of a seat that needs to be around for
2611 the seat to be considered available, usually the graphics
2614 * tmpfiles gained a new "X" line type, that allows
2615 configuration of files and directories (with wildcards) that
2616 shall be excluded from automatic cleanup ("aging").
2618 * udev default rules set the device node permissions now only
2619 at "add" events, and do not change them any longer with a
2620 later "change" event.
2622 * The log messages for lid events and power/sleep keypresses
2623 now carry a message ID.
2625 * We now have a substantially larger unit test suite, but this
2626 continues to be work in progress.
2628 * udevadm hwdb gained a new --root= parameter to change the
2629 root directory to operate relative to.
2631 * logind will now issue a background sync() request to the kernel
2632 early at shutdown, so that dirty buffers are flushed to disk early
2633 instead of at the last moment, in order to optimize shutdown
2636 * A new bootctl tool has been added that is an interface for
2637 certain boot loader operations. This is currently a preview
2638 and is likely to be extended into a small mechanism daemon
2639 like timedated, localed, hostnamed, and can be used by
2640 graphical UIs to enumerate available boot options, and
2641 request boot into firmware operations.
2643 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2644 the rest of the package. It also has been updated to work
2645 correctly in initrds.
2647 * Policykit previously has been runtime optional, and is now
2648 also compile time optional via a configure switch.
2650 * systemd-analyze has been reimplemented in C. Also "systemctl
2651 dot" has moved into systemd-analyze.
2653 * "systemctl status" with no further parameters will now print
2654 the status of all active or failed units.
2656 * Operations such as "systemctl start" can now be executed
2657 with a new mode "--irreversible" which may be used to queue
2658 operations that cannot accidentally be reversed by a later
2659 job queuing. This is by default used to make shutdown
2660 requests more robust.
2662 * The Python API of systemd now gained a new module for
2663 reading journal files.
2665 * A new tool kernel-install has been added that can install
2666 kernel images according to the Boot Loader Specification:
2668 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2670 * Boot time console output has been improved to provide
2671 animated boot time output for hanging jobs.
2673 * A new tool systemd-activate has been added which can be used
2674 to test socket activation with, directly from the command
2675 line. This should make it much easier to test and debug
2676 socket activation in daemons.
2678 * journalctl gained a new "--reverse" (or -r) option to show
2679 journal output in reverse order (i.e. newest line first).
2681 * journalctl gained a new "--pager-end" (or -e) option to jump
2682 to immediately jump to the end of the journal in the
2683 pager. This is only supported in conjunction with "less".
2685 * journalctl gained a new "--user-unit=" option, that works
2686 similarly to "--unit=" but filters for user units rather than
2689 * A number of unit files to ease adoption of systemd in
2690 initrds has been added. This moves some minimal logic from
2691 the various initrd implementations into systemd proper.
2693 * The journal files are now owned by a new group
2694 "systemd-journal", which exists specifically to allow access
2695 to the journal, and nothing else. Previously, we used the
2696 "adm" group for that, which however possibly covers more
2697 than just journal/log file access. This new group is now
2698 already used by systemd-journal-gatewayd to ensure this
2699 daemon gets access to the journal files and as little else
2700 as possible. Note that "make install" will also set FS ACLs
2701 up for /var/log/journal to give "adm" and "wheel" read
2702 access to it, in addition to "systemd-journal" which owns
2703 the journal files. We recommend that packaging scripts also
2704 add read access to "adm" + "wheel" to /var/log/journal, and
2705 all existing/future journal files. To normal users and
2706 administrators little changes, however packagers need to
2707 ensure to create the "systemd-journal" system group at
2708 package installation time.
2710 * The systemd-journal-gatewayd now runs as unprivileged user
2711 systemd-journal-gateway:systemd-journal-gateway. Packaging
2712 scripts need to create these system user/group at
2715 * timedated now exposes a new boolean property CanNTP that
2716 indicates whether a local NTP service is available or not.
2718 * systemd-detect-virt will now also detect xen PVs
2720 * The pstore file system is now mounted by default, if it is
2723 * In addition to the SELinux and IMA policies we will now also
2724 load SMACK policies at early boot.
2726 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2727 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2728 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2729 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2730 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2731 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2732 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2733 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2734 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2735 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2736 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2737 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2738 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2739 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2743 * Timer units now support calendar time events in addition to
2744 monotonic time events. That means you can now trigger a unit
2745 based on a calendar time specification such as "Thu,Fri
2746 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2747 or fifth day of any month of the year 2013, given that it is
2748 a thursday or friday. This brings timer event support
2749 considerably closer to cron's capabilities. For details on
2750 the supported calendar time specification language see
2753 * udev now supports a number of different naming policies for
2754 network interfaces for predictable names, and a combination
2755 of these policies is now the default. Please see this wiki
2756 document for details:
2758 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2760 * Auke Kok's bootchart implementation has been added to the
2761 systemd tree. It is an optional component that can graph the
2762 boot in quite some detail. It is one of the best bootchart
2763 implementations around and minimal in its code and
2766 * nss-myhostname has been integrated into the systemd source
2767 tree. nss-myhostname guarantees that the local hostname
2768 always stays resolvable via NSS. It has been a weak
2769 requirement of systemd-hostnamed since a long time, and
2770 since its code is actually trivial we decided to just
2771 include it in systemd's source tree. It can be turned off
2772 with a configure switch.
2774 * The read-ahead logic is now capable of properly detecting
2775 whether a btrfs file system is on SSD or rotating media, in
2776 order to optimize the read-ahead scheme. Previously, it was
2777 only capable of detecting this on traditional file systems
2780 * In udev, additional device properties are now read from the
2781 IAB in addition to the OUI database. Also, Bluetooth company
2782 identities are attached to the devices as well.
2784 * In service files %U may be used as specifier that is
2785 replaced by the configured user name of the service.
2787 * nspawn may now be invoked without a controlling TTY. This
2788 makes it suitable for invocation as its own service. This
2789 may be used to set up a simple containerized server system
2790 using only core OS tools.
2792 * systemd and nspawn can now accept socket file descriptors
2793 when they are started for socket activation. This enables
2794 implementation of socket activated nspawn
2795 containers. i.e. think about autospawning an entire OS image
2796 when the first SSH or HTTP connection is received. We expect
2797 that similar functionality will also be added to libvirt-lxc
2800 * journalctl will now suppress ANSI color codes when
2801 presenting log data.
2803 * systemctl will no longer show control group information for
2804 a unit if a the control group is empty anyway.
2806 * logind can now automatically suspend/hibernate/shutdown the
2809 * /etc/machine-info and hostnamed now also expose the chassis
2810 type of the system. This can be used to determine whether
2811 the local system is a laptop, desktop, handset or
2812 tablet. This information may either be configured by the
2813 user/vendor or is automatically determined from ACPI and DMI
2814 information if possible.
2816 * A number of PolicyKit actions are now bound together with
2817 "imply" rules. This should simplify creating UIs because
2818 many actions will now authenticate similar ones as well.
2820 * Unit files learnt a new condition ConditionACPower= which
2821 may be used to conditionalize a unit depending on whether an
2822 AC power source is connected or not, of whether the system
2823 is running on battery power.
2825 * systemctl gained a new "is-failed" verb that may be used in
2826 shell scripts and suchlike to check whether a specific unit
2827 is in the "failed" state.
2829 * The EnvironmentFile= setting in unit files now supports file
2830 globbing, and can hence be used to easily read a number of
2831 environment files at once.
2833 * systemd will no longer detect and recognize specific
2834 distributions. All distribution-specific #ifdeffery has been
2835 removed, systemd is now fully generic and
2836 distribution-agnostic. Effectively, not too much is lost as
2837 a lot of the code is still accessible via explicit configure
2838 switches. However, support for some distribution specific
2839 legacy configuration file formats has been dropped. We
2840 recommend distributions to simply adopt the configuration
2841 files everybody else uses now and convert the old
2842 configuration from packaging scripts. Most distributions
2843 already did that. If that's not possible or desirable,
2844 distributions are welcome to forward port the specific
2845 pieces of code locally from the git history.
2847 * When logging a message about a unit systemd will now always
2848 log the unit name in the message meta data.
2850 * localectl will now also discover system locale data that is
2851 not stored in locale archives, but directly unpacked.
2853 * logind will no longer unconditionally use framebuffer
2854 devices as seat masters, i.e. as devices that are required
2855 to be existing before a seat is considered preset. Instead,
2856 it will now look for all devices that are tagged as
2857 "seat-master" in udev. By default framebuffer devices will
2858 be marked as such, but depending on local systems other
2859 devices might be marked as well. This may be used to
2860 integrate graphics cards using closed source drivers (such
2861 as NVidia ones) more nicely into logind. Note however, that
2862 we recommend using the open source NVidia drivers instead,
2863 and no udev rules for the closed-source drivers will be
2864 shipped from us upstream.
2866 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2867 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2868 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2869 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2870 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2871 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2872 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2873 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2874 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2875 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2876 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2877 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2882 * udev gained support for loading additional device properties
2883 from an indexed database that is keyed by vendor/product IDs
2884 and similar device identifiers. For the beginning this
2885 "hwdb" is populated with data from the well-known PCI and
2886 USB database, but also includes PNP, ACPI and OID data. In
2887 the longer run this indexed database shall grow into
2888 becoming the one central database for non-essential
2889 userspace device metadata. Previously, data from the PCI/USB
2890 database was only attached to select devices, since the
2891 lookup was a relatively expensive operation due to O(n) time
2892 complexity (with n being the number of entries in the
2893 database). Since this is now O(1), we decided to add in this
2894 data for all devices where this is available, by
2895 default. Note that the indexed database needs to be rebuilt
2896 when new data files are installed. To achieve this you need
2897 to update your packaging scripts to invoke "udevadm hwdb
2898 --update" after installation of hwdb data files. For
2899 RPM-based distributions we introduced the new
2900 %udev_hwdb_update macro for this purpose.
2902 * The Journal gained support for the "Message Catalog", an
2903 indexed database to link up additional information with
2904 journal entries. For further details please check:
2906 http://www.freedesktop.org/wiki/Software/systemd/catalog
2908 The indexed message catalog database also needs to be
2909 rebuilt after installation of message catalog files. Use
2910 "journalctl --update-catalog" for this. For RPM-based
2911 distributions we introduced the %journal_catalog_update
2912 macro for this purpose.
2914 * The Python Journal bindings gained support for the standard
2915 Python logging framework.
2917 * The Journal API gained new functions for checking whether
2918 the underlying file system of a journal file is capable of
2919 properly reporting file change notifications, or whether
2920 applications that want to reflect journal changes "live"
2921 need to recheck journal files continuously in appropriate
2924 * It is now possible to set the "age" field for tmpfiles
2925 entries to 0, indicating that files matching this entry
2926 shall always be removed when the directories are cleaned up.
2928 * coredumpctl gained a new "gdb" verb which invokes gdb
2929 right-away on the selected coredump.
2931 * There's now support for "hybrid sleep" on kernels that
2932 support this, in addition to "suspend" and "hibernate". Use
2933 "systemctl hybrid-sleep" to make use of this.
2935 * logind's HandleSuspendKey= setting (and related settings)
2936 now gained support for a new "lock" setting to simply
2937 request the screen lock on all local sessions, instead of
2938 actually executing a suspend or hibernation.
2940 * systemd will now mount the EFI variables file system by
2943 * Socket units now gained support for configuration of the
2944 SMACK security label.
2946 * timedatectl will now output the time of the last and next
2947 daylight saving change.
2949 * We dropped support for various legacy and distro-specific
2950 concepts, such as insserv, early-boot SysV services
2951 (i.e. those for non-standard runlevels such as 'b' or 'S')
2952 or ArchLinux /etc/rc.conf support. We recommend the
2953 distributions who still need support this to either continue
2954 to maintain the necessary patches downstream, or find a
2955 different solution. (Talk to us if you have questions!)
2957 * Various systemd components will now bypass PolicyKit checks
2958 for root and otherwise handle properly if PolicyKit is not
2959 found to be around. This should fix most issues for
2960 PolicyKit-less systems. Quite frankly this should have been
2961 this way since day one. It is absolutely our intention to
2962 make systemd work fine on PolicyKit-less systems, and we
2963 consider it a bug if something does not work as it should if
2964 PolicyKit is not around.
2966 * For embedded systems it is now possible to build udev and
2967 systemd without blkid and/or kmod support.
2969 * "systemctl switch-root" is now capable of switching root
2970 more than once. I.e. in addition to transitions from the
2971 initrd to the host OS it is now possible to transition to
2972 further OS images from the host. This is useful to implement
2973 offline updating tools.
2975 * Various other additions have been made to the RPM macros
2976 shipped with systemd. Use %udev_rules_update() after
2977 installing new udev rules files. %_udevhwdbdir,
2978 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
2979 %_sysctldir are now available which resolve to the right
2980 directories for packages to place various data files in.
2982 * journalctl gained the new --full switch (in addition to
2983 --all, to disable ellipsation for long messages.
2985 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
2986 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
2987 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
2988 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
2989 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
2990 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
2991 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
2992 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
2993 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
2997 * journalctl gained new --since= and --until= switches to
2998 filter by time. It also now supports nice filtering for
2999 units via --unit=/-u.
3001 * Type=oneshot services may use ExecReload= and do the
3004 * The journal daemon now supports time-based rotation and
3005 vacuuming, in addition to the usual disk-space based
3008 * The journal will now index the available field values for
3009 each field name. This enables clients to show pretty drop
3010 downs of available match values when filtering. The bash
3011 completion of journalctl has been updated
3012 accordingly. journalctl gained a new switch -F to list all
3013 values a certain field takes in the journal database.
3015 * More service events are now written as structured messages
3016 to the journal, and made recognizable via message IDs.
3018 * The timedated, localed and hostnamed mini-services which
3019 previously only provided support for changing time, locale
3020 and hostname settings from graphical DEs such as GNOME now
3021 also have a minimal (but very useful) text-based client
3022 utility each. This is probably the nicest way to changing
3023 these settings from the command line now, especially since
3024 it lists available options and is fully integrated with bash
3027 * There's now a new tool "systemd-coredumpctl" to list and
3028 extract coredumps from the journal.
3030 * We now install a README each in /var/log/ and
3031 /etc/rc.d/init.d explaining where the system logs and init
3032 scripts went. This hopefully should help folks who go to
3033 that dirs and look into the otherwise now empty void and
3034 scratch their heads.
3036 * When user-services are invoked (by systemd --user) the
3037 $MANAGERPID env var is set to the PID of systemd.
3039 * SIGRTMIN+24 when sent to a --user instance will now result
3040 in immediate termination of systemd.
3042 * gatewayd received numerous feature additions such as a
3043 "follow" mode, for live syncing and filtering.
3045 * browse.html now allows filtering and showing detailed
3046 information on specific entries. Keyboard navigation and
3047 mouse screen support has been added.
3049 * gatewayd/journalctl now supports HTML5/JSON
3050 Server-Sent-Events as output.
3052 * The SysV init script compatibility logic will now
3053 heuristically determine whether a script supports the
3054 "reload" verb, and only then make this available as
3057 * "systemctl status --follow" has been removed, use "journalctl
3060 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3061 have been removed since they are hardly useful to be
3064 * And I'd like to take the opportunity to specifically mention
3065 Zbigniew for his great contributions. Zbigniew, you rock!
3067 Contributions from: Andrew Eikum, Christian Hesse, Colin
3068 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3069 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3070 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3071 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3072 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3073 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3077 * If /etc/vconsole.conf is non-existent or empty we will no
3078 longer load any console font or key map at boot by
3079 default. Instead the kernel defaults will be left
3080 intact. This is definitely the right thing to do, as no
3081 configuration should mean no configuration, and hard-coding
3082 font names that are different on all archs is probably a bad
3083 idea. Also, the kernel default key map and font should be
3084 good enough for most cases anyway, and mostly identical to
3085 the userspace fonts/key maps we previously overloaded them
3086 with. If distributions want to continue to default to a
3087 non-kernel font or key map they should ship a default
3088 /etc/vconsole.conf with the appropriate contents.
3090 Contributions from: Colin Walters, Daniel J Walsh, Dave
3091 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3092 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3096 * journalctl gained a new --cursor= switch to show entries
3097 starting from the specified location in the journal.
3099 * We now enforce a size limit on journal entry fields exported
3100 with "-o json" in journalctl. Fields larger than 4K will be
3101 assigned null. This can be turned off with --all.
3103 * An (optional) journal gateway daemon is now available as
3104 "systemd-journal-gatewayd.service". This service provides
3105 access to the journal via HTTP and JSON. This functionality
3106 will be used to implement live log synchronization in both
3107 pull and push modes, but has various other users too, such
3108 as easy log access for debugging of embedded devices. Right
3109 now it is already useful to retrieve the journal via HTTP:
3111 # systemctl start systemd-journal-gatewayd.service
3112 # wget http://localhost:19531/entries
3114 This will download the journal contents in a
3115 /var/log/messages compatible format. The same as JSON:
3117 # curl -H"Accept: application/json" http://localhost:19531/entries
3119 This service is also accessible via a web browser where a
3120 single static HTML5 app is served that uses the JSON logic
3121 to enable the user to do some basic browsing of the
3122 journal. This will be extended later on. Here's an example
3123 screenshot of this app in its current state:
3125 http://0pointer.de/public/journal-gatewayd
3127 Contributions from: Kay Sievers, Lennart Poettering, Robert
3128 Milasan, Tom Gundersen
3132 * The bash completion logic is now available for journalctl
3135 * We do not mount the "cpuset" controller anymore together with
3136 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
3137 started if no parameters are assigned to it. "cpuset" hence
3138 broke code that assumed it it could create "cpu" groups and
3141 * journalctl -f will now subscribe to terminal size changes,
3142 and line break accordingly.
3144 Contributions from: Dave Reisner, Kay Sievers, Lennart
3145 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3149 * nspawn will now create a symlink /etc/localtime in the
3150 container environment, copying the host's timezone
3151 setting. Previously this has been done via a bind mount, but
3152 since symlinks cannot be bind mounted this has now been
3153 changed to create/update the appropriate symlink.
3155 * journalctl -n's line number argument is now optional, and
3156 will default to 10 if omitted.
3158 * journald will now log the maximum size the journal files may
3159 take up on disk. This is particularly useful if the default
3160 built-in logic of determining this parameter from the file
3161 system size is used. Use "systemctl status
3162 systemd-journald.service" to see this information.
3164 * The multi-seat X wrapper tool has been stripped down. As X
3165 is now capable of enumerating graphics devices via udev in a
3166 seat-aware way the wrapper is not strictly necessary
3167 anymore. A stripped down temporary stop-gap is still shipped
3168 until the upstream display managers have been updated to
3169 fully support the new X logic. Expect this wrapper to be
3170 removed entirely in one of the next releases.
3172 * HandleSleepKey= in logind.conf has been split up into
3173 HandleSuspendKey= and HandleHibernateKey=. The old setting
3174 is not available anymore. X11 and the kernel are
3175 distuingishing between these keys and we should too. This
3176 also means the inhibition lock for these keys has been split
3179 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
3180 Poettering, Lukas Nykryn, Václav Pavlín
3184 * Whenever a unit changes state we will now log this to the
3185 journal and show along the unit's own log output in
3188 * ConditionPathIsMountPoint= can now properly detect bind
3189 mount points too. (Previously, a bind mount of one file
3190 system to another place in the same file system could not be
3191 detected as mount, since they shared struct stat's st_dev
3194 * We will now mount the cgroup controllers cpu, cpuacct,
3195 cpuset and the controllers net_cls, net_prio together by
3198 * nspawn containers will now have a virtualized boot
3199 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
3200 over with a randomized ID at container initialization). This
3201 has the effect of making "journalctl -b" do the right thing
3204 * The JSON output journal serialization has been updated not
3205 to generate "endless" list objects anymore, but rather one
3206 JSON object per line. This is more in line how most JSON
3207 parsers expect JSON objects. The new output mode
3208 "json-pretty" has been added to provide similar output, but
3209 neatly aligned for readability by humans.
3211 * We dropped all explicit sync() invocations in the shutdown
3212 code. The kernel does this implicitly anyway in the kernel
3213 reboot() syscall. halt(8)'s -n option is now a compatibility
3216 * We now support virtualized reboot() in containers, as
3217 supported by newer kernels. We will fall back to exit() if
3218 CAP_SYS_REBOOT is not available to the container. Also,
3219 nspawn makes use of this now and will actually reboot the
3220 container if the containerized OS asks for that.
3222 * journalctl will only show local log output by default
3223 now. Use --merge (-m) to show remote log output, too.
3225 * libsystemd-journal gained the new sd_journal_get_usage()
3226 call to determine the current disk usage of all journal
3227 files. This is exposed in the new "journalctl --disk-usage"
3230 * journald gained a new configuration setting SplitMode= in
3231 journald.conf which may be used to control how user journals
3232 are split off. See journald.conf(5) for details.
3234 * A new condition type ConditionFileNotEmpty= has been added.
3236 * tmpfiles' "w" lines now support file globbing, to write
3237 multiple files at once.
3239 * We added Python bindings for the journal submission
3240 APIs. More Python APIs for a number of selected APIs will
3241 likely follow. Note that we intend to add native bindings
3242 only for the Python language, as we consider it common
3243 enough to deserve bindings shipped within systemd. There are
3244 various projects outside of systemd that provide bindings
3245 for languages such as PHP or Lua.
3247 * Many conditions will now resolve specifiers such as %i. In
3248 addition, PathChanged= and related directives of .path units
3249 now support specifiers as well.
3251 * There's now a new RPM macro definition for the system preset
3254 * journald will now warn if it ca not forward a message to the
3255 syslog daemon because its socket is full.
3257 * timedated will no longer write or process /etc/timezone,
3258 except on Debian. As we do not support late mounted /usr
3259 anymore /etc/localtime always being a symlink is now safe,
3260 and hence the information in /etc/timezone is not necessary
3263 * logind will now always reserve one VT for a text getty (VT6
3264 by default). Previously if more than 6 X sessions where
3265 started they took up all the VTs with auto-spawned gettys,
3266 so that no text gettys were available anymore.
3268 * udev will now automatically inform the btrfs kernel logic
3269 about btrfs RAID components showing up. This should make
3270 simple hotplug based btrfs RAID assembly work.
3272 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
3273 (but not for its children which will stay at the kernel
3274 default). This should allow setups with a lot more listening
3277 * systemd will now always pass the configured timezone to the
3278 kernel at boot. timedated will do the same when the timezone
3281 * logind's inhibition logic has been updated. By default,
3282 logind will now handle the lid switch, the power and sleep
3283 keys all the time, even in graphical sessions. If DEs want
3284 to handle these events on their own they should take the new
3285 handle-power-key, handle-sleep-key and handle-lid-switch
3286 inhibitors during their runtime. A simple way to achiveve
3287 that is to invoke the DE wrapped in an invocation of:
3289 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
3291 * Access to unit operations is now checked via SELinux taking
3292 the unit file label and client process label into account.
3294 * systemd will now notify the administrator in the journal
3295 when he over-mounts a non-empty directory.
3297 * There are new specifiers that are resolved in unit files,
3298 for the host name (%H), the machine ID (%m) and the boot ID
3301 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
3302 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
3303 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
3304 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
3305 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
3306 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3307 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
3311 * Support for reading structured kernel messages from
3312 /dev/kmsg has now been added and is enabled by default.
3314 * Support for reading kernel messages from /proc/kmsg has now
3315 been removed. If you want kernel messages in the journal
3316 make sure to run a recent kernel (>= 3.5) that supports
3317 reading structured messages from /dev/kmsg (see
3318 above). /proc/kmsg is now exclusive property of classic
3319 syslog daemons again.
3321 * The libudev API gained the new
3322 udev_device_new_from_device_id() call.
3324 * The logic for file system namespace (ReadOnlyDirectory=,
3325 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
3326 require pivot_root() anymore. This means fewer temporary
3327 directories are created below /tmp for this feature.
3329 * nspawn containers will now see and receive all submounts
3330 made on the host OS below the root file system of the
3333 * Forward Secure Sealing is now supported for Journal files,
3334 which provide cryptographical sealing of journal files so
3335 that attackers cannot alter log history anymore without this
3336 being detectable. Lennart will soon post a blog story about
3337 this explaining it in more detail.
3339 * There are two new service settings RestartPreventExitStatus=
3340 and SuccessExitStatus= which allow configuration of exit
3341 status (exit code or signal) which will be excepted from the
3342 restart logic, resp. consider successful.
3344 * journalctl gained the new --verify switch that can be used
3345 to check the integrity of the structure of journal files and
3346 (if Forward Secure Sealing is enabled) the contents of
3349 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3350 and similar symlinks pre-created. This makes running shells
3351 as container init process a lot more fun.
3353 * The fstab support can now handle PARTUUID= and PARTLABEL=
3356 * A new ConditionHost= condition has been added to match
3357 against the hostname (with globs) and machine ID. This is
3358 useful for clusters where a single OS image is used to
3359 provision a large number of hosts which shall run slightly
3360 different sets of services.
3362 * Services which hit the restart limit will now be placed in a
3365 Contributions from: Bertram Poettering, Dave Reisner, Huang
3366 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3367 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3371 * When running in --user mode systemd will now become a
3372 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3373 tree a lot more organized.
3375 * A new PartOf= unit dependency type has been introduced that
3376 may be used to group services in a natural way.
3378 * "systemctl enable" may now be used to enable instances of
3381 * journalctl now prints error log levels in red, and
3382 warning/notice log levels in bright white. It also supports
3383 filtering by log level now.
3385 * cgtop gained a new -n switch (similar to top), to configure
3386 the maximum number of iterations to run for. It also gained
3387 -b, to run in batch mode (accepting no input).
3389 * The suffix ".service" may now be omitted on most systemctl
3390 command lines involving service unit names.
3392 * There's a new bus call in logind to lock all sessions, as
3393 well as a loginctl verb for it "lock-sessions".
3395 * libsystemd-logind.so gained a new call sd_journal_perror()
3396 that works similar to libc perror() but logs to the journal
3397 and encodes structured information about the error number.
3399 * /etc/crypttab entries now understand the new keyfile-size=
3402 * shutdown(8) now can send a (configurable) wall message when
3403 a shutdown is cancelled.
3405 * The mount propagation mode for the root file system will now
3406 default to "shared", which is useful to make containers work
3407 nicely out-of-the-box so that they receive new mounts from
3408 the host. This can be undone locally by running "mount
3409 --make-rprivate /" if needed.
3411 * The prefdm.service file has been removed. Distributions
3412 should maintain this unit downstream if they intend to keep
3413 it around. However, we recommend writing normal unit files
3414 for display managers instead.
3416 * Since systemd is a crucial part of the OS we will now
3417 default to a number of compiler switches that improve
3418 security (hardening) such as read-only relocations, stack
3419 protection, and suchlike.
3421 * The TimeoutSec= setting for services is now split into
3422 TimeoutStartSec= and TimeoutStopSec= to allow configuration
3423 of individual time outs for the start and the stop phase of
3426 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
3427 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
3428 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
3429 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
3430 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
3431 Gundersen, Zbigniew Jędrzejewski-Szmek
3435 * The journal and id128 C APIs are now fully documented as man
3438 * Extra safety checks have been added when transitioning from
3439 the initial RAM disk to the main system to avoid accidental
3442 * /etc/crypttab entries now understand the new keyfile-offset=
3445 * systemctl -t can now be used to filter by unit load state.
3447 * The journal C API gained the new sd_journal_wait() call to
3448 make writing synchronous journal clients easier.
3450 * journalctl gained the new -D switch to show journals from a
3453 * journalctl now displays a special marker between log
3454 messages of two different boots.
3456 * The journal is now explicitly flushed to /var via a service
3457 systemd-journal-flush.service, rather than implicitly simply
3458 by seeing /var/log/journal to be writable.
3460 * journalctl (and the journal C APIs) can now match for much
3461 more complex expressions, with alternatives and
3464 * When transitioning from the initial RAM disk to the main
3465 system we will now kill all processes in a killing spree to
3466 ensure no processes stay around by accident.
3468 * Three new specifiers may be used in unit files: %u, %h, %s
3469 resolve to the user name, user home directory resp. user
3470 shell. This is useful for running systemd user instances.
3472 * We now automatically rotate journal files if their data
3473 object hash table gets a fill level > 75%. We also size the
3474 hash table based on the configured maximum file size. This
3475 together should lower hash collisions drastically and thus
3476 speed things up a bit.
3478 * journalctl gained the new "--header" switch to introspect
3479 header data of journal files.
3481 * A new setting SystemCallFilters= has been added to services
3482 which may be used to apply blacklists or whitelists to
3483 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
3485 * nspawn gained a new --link-journal= switch (and quicker: -j)
3486 to link the container journal with the host. This makes it
3487 very easy to centralize log viewing on the host for all
3488 guests while still keeping the journal files separated.
3490 * Many bugfixes and optimizations
3492 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
3493 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
3494 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
3499 * Several tools now understand kernel command line arguments,
3500 which are only read when run in an initial RAM disk. They
3501 usually follow closely their normal counterparts, but are
3504 * There's a new tool to analyze the readahead files that are
3505 automatically generated at boot. Use:
3507 /usr/lib/systemd/systemd-readahead analyze /.readahead
3509 * We now provide an early debug shell on tty9 if this enabled. Use:
3511 systemctl enable debug-shell.service
3513 * All plymouth related units have been moved into the Plymouth
3514 package. Please make sure to upgrade your Plymouth version
3517 * systemd-tmpfiles now supports getting passed the basename of
3518 a configuration file only, in which case it will look for it
3519 in all appropriate directories automatically.
3521 * udevadm info now takes a /dev or /sys path as argument, and
3522 does the right thing. Example:
3524 udevadm info /dev/sda
3525 udevadm info /sys/class/block/sda
3527 * systemctl now prints a warning if a unit is stopped but a
3528 unit that might trigger it continues to run. Example: a
3529 service is stopped but the socket that activates it is left
3532 * "systemctl status" will now mention if the log output was
3533 shortened due to rotation since a service has been started.
3535 * The journal API now exposes functions to determine the
3536 "cutoff" times due to rotation.
3538 * journald now understands SIGUSR1 and SIGUSR2 for triggering
3539 immediately flushing of runtime logs to /var if possible,
3540 resp. for triggering immediate rotation of the journal
3543 * It is now considered an error if a service is attempted to
3544 be stopped that is not loaded.
3546 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
3548 * systemd-analyze now supports Python 3
3550 * tmpfiles now supports cleaning up directories via aging
3551 where the first level dirs are always kept around but
3552 directories beneath it automatically aged. This is enabled
3553 by prefixing the age field with '~'.
3555 * Seat objects now expose CanGraphical, CanTTY properties
3556 which is required to deal with very fast bootups where the
3557 display manager might be running before the graphics drivers
3558 completed initialization.
3560 * Seat objects now expose a State property.
3562 * We now include RPM macros for service enabling/disabling
3563 based on the preset logic. We recommend RPM based
3564 distributions to make use of these macros if possible. This
3565 makes it simpler to reuse RPM spec files across
3568 * We now make sure that the collected systemd unit name is
3569 always valid when services log to the journal via
3572 * There's a new man page kernel-command-line(7) detailing all
3573 command line options we understand.
3575 * The fstab generator may now be disabled at boot by passing
3576 fstab=0 on the kernel command line.
3578 * A new kernel command line option modules-load= is now understood
3579 to load a specific kernel module statically, early at boot.
3581 * Unit names specified on the systemctl command line are now
3582 automatically escaped as needed. Also, if file system or
3583 device paths are specified they are automatically turned
3584 into the appropriate mount or device unit names. Example:
3586 systemctl status /home
3587 systemctl status /dev/sda
3589 * The SysVConsole= configuration option has been removed from
3590 system.conf parsing.
3592 * The SysV search path is no longer exported on the D-Bus
3595 * The Names= option is been removed from unit file parsing.
3597 * There's a new man page bootup(7) detailing the boot process.
3599 * Every unit and every generator we ship with systemd now
3600 comes with full documentation. The self-explanatory boot is
3603 * A couple of services gained "systemd-" prefixes in their
3604 name if they wrap systemd code, rather than only external
3605 code. Among them fsck@.service which is now
3606 systemd-fsck@.service.
3608 * The HaveWatchdog property has been removed from the D-Bus
3611 * systemd.confirm_spawn= on the kernel command line should now
3614 * There's a new man page crypttab(5) which details all options
3615 we actually understand.
3617 * systemd-nspawn gained a new --capability= switch to pass
3618 additional capabilities to the container.
3620 * timedated will now read known NTP implementation unit names
3621 from /usr/lib/systemd/ntp-units.d/*.list,
3622 systemd-timedated-ntp.target has been removed.
3624 * journalctl gained a new switch "-b" that lists log data of
3625 the current boot only.
3627 * The notify socket is in the abstract namespace again, in
3628 order to support daemons which chroot() at start-up.
3630 * There is a new Storage= configuration option for journald
3631 which allows configuration of where log data should go. This
3632 also provides a way to disable journal logging entirely, so
3633 that data collected is only forwarded to the console, the
3634 kernel log buffer or another syslog implementation.
3636 * Many bugfixes and optimizations
3638 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3639 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3640 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3641 Shawn Landden, Tom Gundersen
3645 * "systemctl help <unit>" now shows the man page if one is
3648 * Several new man pages have been added.
3650 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3651 MaxLevelConsole= can now be specified in
3652 journald.conf. These options allow reducing the amount of
3653 data stored on disk or forwarded by the log level.
3655 * TimerSlackNSec= can now be specified in system.conf for
3656 PID1. This allows system-wide power savings.
3658 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3659 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3664 * logind is now capable of (optionally) handling power and
3665 sleep keys as well as the lid switch.
3667 * journalctl now understands the syntax "journalctl
3668 /usr/bin/avahi-daemon" to get all log output of a specific
3671 * CapabilityBoundingSet= in system.conf now also influences
3672 the capability bound set of usermode helpers of the kernel.
3674 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3675 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3676 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3677 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3681 * Note that we skipped 139 releases here in order to set the
3682 new version to something that is greater than both udev's
3683 and systemd's most recent version number.
3685 * udev: all udev sources are merged into the systemd source tree now.
3686 All future udev development will happen in the systemd tree. It
3687 is still fully supported to use the udev daemon and tools without
3688 systemd running, like in initramfs or other init systems. Building
3689 udev though, will require the *build* of the systemd tree, but
3690 udev can be properly *run* without systemd.
3692 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3693 should be used to create dead device nodes as workarounds for broken
3696 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3697 no longer supported. udev_monitor_new_from_netlink() needs to be
3698 used to subscribe to events.
3700 * udev: when udevd is started by systemd, processes which are left
3701 behind by forking them off of udev rules, are unconditionally cleaned
3702 up and killed now after the event handling has finished. Services or
3703 daemons must be started as systemd services. Services can be
3704 pulled-in by udev to get started, but they can no longer be directly
3705 forked by udev rules.
3707 * udev: the daemon binary is called systemd-udevd now and installed
3708 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3709 to adapt to that, create symlink, or rename the binary after building
3712 * libudev no longer provides these symbols:
3713 udev_monitor_from_socket()
3714 udev_queue_get_failed_list_entry()
3715 udev_get_{dev,sys,run}_path()
3716 The versions number was bumped and symbol versioning introduced.
3718 * systemd-loginctl and systemd-journalctl have been renamed
3719 to loginctl and journalctl to match systemctl.
3721 * The config files: /etc/systemd/systemd-logind.conf and
3722 /etc/systemd/systemd-journald.conf have been renamed to
3723 logind.conf and journald.conf. Package updates should rename
3724 the files to the new names on upgrade.
3726 * For almost all files the license is now LGPL2.1+, changed
3727 from the previous GPL2.0+. Exceptions are some minor stuff
3728 of udev (which will be changed to LGPL2.1 eventually, too),
3729 and the MIT licensed sd-daemon.[ch] library that is suitable
3730 to be used as drop-in files.
3732 * systemd and logind now handle system sleep states, in
3733 particular suspending and hibernating.
3735 * logind now implements a sleep/shutdown/idle inhibiting logic
3736 suitable for a variety of uses. Soonishly Lennart will blog
3737 about this in more detail.
3739 * var-run.mount and var-lock.mount are no longer provided
3740 (which prevously bind mounted these directories to their new
3741 places). Distributions which have not converted these
3742 directories to symlinks should consider stealing these files
3743 from git history and add them downstream.
3745 * We introduced the Documentation= field for units and added
3746 this to all our shipped units. This is useful to make it
3747 easier to explore the boot and the purpose of the various
3750 * All smaller setup units (such as
3751 systemd-vconsole-setup.service) now detect properly if they
3752 are run in a container and are skipped when
3753 appropriate. This guarantees an entirely noise-free boot in
3754 Linux container environments such as systemd-nspawn.
3756 * A framework for implementing offline system updates is now
3757 integrated, for details see:
3758 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3760 * A new service type Type=idle is available now which helps us
3761 avoiding ugly interleaving of getty output and boot status
3764 * There's now a system-wide CapabilityBoundingSet= option to
3765 globally reduce the set of capabilities for the
3766 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3767 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3768 even CAP_NET_ADMIN system-wide for secure systems.
3770 * There are now system-wide DefaultLimitXXX= options to
3771 globally change the defaults of the various resource limits
3772 for all units started by PID 1.
3774 * Harald Hoyer's systemd test suite has been integrated into
3775 systemd which allows easy testing of systemd builds in qemu
3776 and nspawn. (This is really awesome! Ask us for details!)
3778 * The fstab parser is now implemented as generator, not inside
3781 * systemctl will now warn you if .mount units generated from
3782 /etc/fstab are out of date due to changes in fstab that
3783 have not been read by systemd yet.
3785 * systemd is now suitable for usage in initrds. Dracut has
3786 already been updated to make use of this. With this in place
3787 initrds get a slight bit faster but primarily are much
3788 easier to introspect and debug since "systemctl status" in
3789 the host system can be used to introspect initrd services,
3790 and the journal from the initrd is kept around too.
3792 * systemd-delta has been added, a tool to explore differences
3793 between user/admin configuration and vendor defaults.
3795 * PrivateTmp= now affects both /tmp and /var/tmp.
3797 * Boot time status messages are now much prettier and feature
3798 proper english language. Booting up systemd has never been
3801 * Read-ahead pack files now include the inode number of all
3802 files to pre-cache. When the inode changes the pre-caching
3803 is not attempted. This should be nicer to deal with updated
3804 packages which might result in changes of read-ahead
3807 * We now temporaritly lower the kernel's read_ahead_kb variable
3808 when collecting read-ahead data to ensure the kernel's
3809 built-in read-ahead does not add noise to our measurements
3810 of necessary blocks to pre-cache.
3812 * There's now RequiresMountsFor= to add automatic dependencies
3813 for all mounts necessary for a specific file system path.
3815 * MountAuto= and SwapAuto= have been removed from
3816 system.conf. Mounting file systems at boot has to take place
3819 * nspawn now learned a new switch --uuid= to set the machine
3820 ID on the command line.
3822 * nspawn now learned the -b switch to automatically search
3825 * vt102 is now the default TERM for serial TTYs, upgraded from
3828 * systemd-logind now works on VT-less systems.
3830 * The build tree has been reorganized. The individual
3831 components now have directories of their own.
3833 * A new condition type ConditionPathIsReadWrite= is now available.
3835 * nspawn learned the new -C switch to create cgroups for the
3836 container in other hierarchies.
3838 * We now have support for hardware watchdogs, configurable in
3841 * The scheduled shutdown logic now has a public API.
3843 * We now mount /tmp as tmpfs by default, but this can be
3844 masked and /etc/fstab can override it.
3846 * Since udisks does not make use of /media anymore we are not
3847 mounting a tmpfs on it anymore.
3849 * journalctl gained a new --local switch to only interleave
3850 locally generated journal files.
3852 * We can now load the IMA policy at boot automatically.
3854 * The GTK tools have been split off into a systemd-ui.
3856 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3857 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3858 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3859 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3860 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3861 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3862 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3863 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3864 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3869 * This is mostly a bugfix release
3871 * Support optional initialization of the machine ID from the
3872 KVM or container configured UUID.
3874 * Support immediate reboots with "systemctl reboot -ff"
3876 * Show /etc/os-release data in systemd-analyze output
3878 * Many bugfixes for the journal, including endianness fixes and
3879 ensuring that disk space enforcement works
3881 * sd-login.h is C++ comptaible again
3883 * Extend the /etc/os-release format on request of the Debian
3886 * We now refuse non-UTF8 strings used in various configuration
3887 and unit files. This is done to ensure we do not pass invalid
3888 data over D-Bus or expose it elsewhere.
3890 * Register Mimo USB Screens as suitable for automatic seat
3893 * Read SELinux client context from journal clients in a race
3896 * Reorder configuration file lookup order. /etc now always
3897 overrides /run in order to allow the administrator to always
3898 and unconditionally override vendor supplied or
3899 automatically generated data.
3901 * The various user visible bits of the journal now have man
3902 pages. We still lack man pages for the journal API calls
3905 * We now ship all man pages in HTML format again in the
3908 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3909 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3910 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3915 * This is mostly a bugfix release
3917 * systems lacking /etc/os-release are no longer supported.
3919 * Various functionality updates to libsystemd-login.so
3921 * Track class of PAM logins to distuingish greeters from
3924 Contributions from: Kay Sievers, Lennart Poettering, Michael
3929 * This is an important bugfix release for v41.
3931 * Building man pages is now optional which should be useful
3932 for those building systemd from git but unwilling to install
3935 * Watchdog support for supervising services is now usable. In
3936 a future release support for hardware watchdogs
3937 (i.e. /dev/watchdog) will be added building on this.
3939 * Service start rate limiting is now configurable and can be
3940 turned off per service. When a start rate limit is hit a
3941 reboot can automatically be triggered.
3943 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3945 Contributions from: Benjamin Franzke, Bill Nottingham,
3946 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3947 Schmidt, Michał Górny, Piotr Drąg
3951 * The systemd binary is installed /usr/lib/systemd/systemd now;
3952 An existing /sbin/init symlink needs to be adapted with the
3955 * The code that loads kernel modules has been ported to invoke
3956 libkmod directly, instead of modprobe. This means we do not
3957 support systems with module-init-tools anymore.
3959 * Watchdog support is now already useful, but still not
3962 * A new kernel command line option systemd.setenv= is
3963 understood to set system wide environment variables
3964 dynamically at boot.
3966 * We now limit the set of capabilities of systemd-journald.
3968 * We now set SIGPIPE to ignore by default, since it only is
3969 useful in shell pipelines, and has little use in general
3970 code. This can be disabled with IgnoreSIPIPE=no in unit
3973 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
3974 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
3979 * This is mostly a bugfix release
3981 * We now expose the reason why a service failed in the
3982 "Result" D-Bus property.
3984 * Rudimentary service watchdog support (will be completed over
3985 the next few releases.)
3987 * When systemd forks off in order execute some service we will
3988 now immediately changes its argv[0] to reflect which process
3989 it will execute. This is useful to minimize the time window
3990 with a generic argv[0], which makes bootcharts more useful
3992 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
3993 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
3994 Mike Kazantsev, Ray Strode
3998 * This is mostly a test release, but incorporates many
4001 * New systemd-cgtop tool to show control groups by their
4004 * Linking against libacl for ACLs is optional again. If
4005 disabled, support tracking device access for active logins
4006 goes becomes unavailable, and so does access to the user
4007 journals by the respective users.
4009 * If a group "adm" exists, journal files are automatically
4010 owned by them, thus allow members of this group full access
4011 to the system journal as well as all user journals.
4013 * The journal now stores the SELinux context of the logging
4014 client for all entries.
4016 * Add C++ inclusion guards to all public headers
4018 * New output mode "cat" in the journal to print only text
4019 messages, without any meta data like date or time.
4021 * Include tiny X server wrapper as a temporary stop-gap to
4022 teach XOrg udev display enumeration. This is used by display
4023 managers such as gdm, and will go away as soon as XOrg
4024 learned native udev hotplugging for display devices.
4026 * Add new systemd-cat tool for executing arbitrary programs
4027 with STDERR/STDOUT connected to the journal. Can also act as
4028 BSD logger replacement, and does so by default.
4030 * Optionally store all locally generated coredumps in the
4031 journal along with meta data.
4033 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
4034 writing short strings to files (for usage for /sys), and for
4035 creating symlinks, character and block device nodes.
4037 * New unit file option ControlGroupPersistent= to make cgroups
4038 persistent, following the mechanisms outlined in
4039 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
4041 * Support multiple local RTCs in a sane way
4043 * No longer monopolize IO when replaying readahead data on
4044 rotating disks, since we might starve non-file-system IO to
4045 death, since fanotify() will not see accesses done by blkid,
4048 * Do not show kernel threads in systemd-cgls anymore, unless
4049 requested with new -k switch.
4051 Contributions from: Dan Horák, Kay Sievers, Lennart
4052 Poettering, Michal Schmidt
4056 * This is mostly a test release, but incorporates many
4059 * The git repository moved to:
4060 git://anongit.freedesktop.org/systemd/systemd
4061 ssh://git.freedesktop.org/git/systemd/systemd
4063 * First release with the journal
4064 http://0pointer.de/blog/projects/the-journal.html
4066 * The journal replaces both systemd-kmsg-syslogd and
4067 systemd-stdout-bridge.
4069 * New sd_pid_get_unit() API call in libsystemd-logind
4071 * Many systemadm clean-ups
4073 * Introduce remote-fs-pre.target which is ordered before all
4074 remote mounts and may be used to start services before all
4077 * Added Mageia support
4079 * Add bash completion for systemd-loginctl
4081 * Actively monitor PID file creation for daemons which exit in
4082 the parent process before having finished writing the PID
4083 file in the daemon process. Daemons which do this need to be
4084 fixed (i.e. PID file creation must have finished before the
4085 parent exits), but we now react a bit more gracefully to them.
4087 * Add colourful boot output, mimicking the well-known output
4088 of existing distributions.
4090 * New option PassCredentials= for socket units, for
4091 compatibility with a recent kernel ABI breakage.
4093 * /etc/rc.local is now hooked in via a generator binary, and
4094 thus will no longer act as synchronization point during
4097 * systemctl list-unit-files now supports --root=.
4099 * systemd-tmpfiles now understands two new commands: z, Z for
4100 relabelling files according to the SELinux database. This is
4101 useful to apply SELinux labels to specific files in /sys,
4104 * Output of SysV services is now forwarded to both the console
4105 and the journal by default, not only just the console.
4107 * New man pages for all APIs from libsystemd-login.
4109 * The build tree got reorganized and a the build system is a
4110 lot more modular allowing embedded setups to specifically
4111 select the components of systemd they are interested in.
4113 * Support for Linux systems lacking the kernel VT subsystem is
4116 * configure's --with-rootdir= got renamed to
4117 --with-rootprefix= to follow the naming used by udev and
4120 * Unless specified otherwise we will now install to /usr instead
4121 of /usr/local by default.
4123 * Processes with '@' in argv[0][0] are now excluded from the
4124 final shut-down killing spree, following the logic explained
4126 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
4128 * All processes remaining in a service cgroup when we enter
4129 the START or START_PRE states are now killed with
4130 SIGKILL. That means it is no longer possible to spawn
4131 background processes from ExecStart= lines (which was never
4132 supported anyway, and bad style).
4134 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
4135 reloading of units together.
4137 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
4138 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
4139 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
4140 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
4141 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek