1 systemd System and Service Manager
5 * Introduce a new API "sd-hwdb.h" for querying the hardware
6 metadata database. With this minimal interface one can query
7 and enumerate the udev hwdb, decoupled from the old libudev
8 library. libudev's interface for this is now only a wrapper
9 around sd-hwdb. A new tool systemd-hwdb has been added to
10 interface with and update the database.
12 * When any of systemd's tools copies files (for example due to
13 tmpfiles' C lines) a btrfs reflink will attempted first,
14 before bytewise copying is done.
16 * systemd-nspawn gained a new --ephemeral switch. When
17 specified a btrfs snapshot is taken of the container's root
18 directory, and immediately removed when the container
19 terminates again. Thus, a container can be started whose
20 changes never alter the container's root directory, and are
21 lost on container termination. This switch can also be used
22 for starting a container off the root file system of the
23 host without affecting the host OS. This switch is only
24 available on btrfs file systems.
26 * systemd-nspawn gained a new --template= switch. It takes the
27 path to a container tree to use as template for the tree
28 specified via --directory=, should that directory be
29 missing. This allows instantiating containers dynamically,
30 on first run. This switch is only available on btrfs file
33 * When a .mount unit refers to a mount point on which multiple
34 mounts are stacked, and the .mount unit is stopped all of
35 the stacked mount points will now be unmounted until no
38 * systemd now has an explicit notion of supported and
39 unsupported unit types. Jobs enqueued for unsupported unit
40 types will now fail with an "unsupported" error code. More
41 specifically .swap, .automount and .device units are not
42 supported in containers, .busname units are not supported on
43 non-kdbus systems. .swap and .automount are also not
44 supported if their respective kernel compile time options
47 * machinectl gained support for two new "copy-from" and
48 "copy-to" commands for copying files from a running
49 container to the host or vice versa.
51 * machinectl gained support for a new "bind" command to bind
52 mount host directories into local containers. This is
53 currently only supported for nspawn containers.
55 * networkd gained support for configuring bridge forwarding
56 database entries (fdb) from .network files.
58 * A new tiny daemon "systemd-importd" has been added that can
59 download container images in tar, raw, qcow2 or dkr formats,
60 and make them available locally in /var/lib/machines, so
61 that they can run as nspawn containers. The daemon can GPG
62 verify the downloads (not supported for dkr, since it has no
63 provisions for verifying downloads). It will transparently
64 decompress bz2, xz, gzip compressed downloads if necessary,
65 and restore sparse files on disk. The daemon uses privilege
66 separation to ensure the actual download logic runs with
67 fewer privileges than the deamon itself. machinectl has
68 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
69 make the functionality of importd available to the
70 user. With this in place the Fedora and Ubuntu "Cloud"
71 images can be downloaded and booted as containers unmodified
72 (the Fedora images lack the appropriate GPG signature files
73 currently, so they cannot be verified, but this will change
74 soon, hopefully). Note that downloading images is currently
75 only fully supported on btrfs.
77 * machinectl is now able to list container images found in
78 /var/lib/machines, along with some metadata about sizes of
79 disk and similar. If the directory is located on btrfs and
80 quota is enabled, this includes quota display. A new command
81 "image-status" has been added that shows additional
82 information about images.
84 * machinectl is now able to clone container images
85 efficiently, if the underlying file system (btrfs) supports
86 it, with the new "machinectl list-images" command. It also
87 gained commands for renaming and removing images, as well as
88 marking them read-only or read-write (supported also on
91 * networkd gained support for collecting LLDP network
92 announcements, from hardware that supports this. This is
93 shown in networkctl output.
95 * systemd-run gained support for a new -t (--pty) switch for
96 invoking a binary on a pty whose input and output is
97 connected to the invoking terminal. This allows executing
98 processes as system services while interactively
99 communicating with them via the terminal. Most interestingly
100 this is supported across container boundaries. Invoking
101 "systemd-run -t /bin/bash" is an alternative to running a
102 full login session, the difference being that the former
103 will not register a session, nor go through the PAM session
106 * tmpfiles gained support for a new "v" line type for creating
107 btrfs subvolumes. If the underlying file system is a legacy
108 file system, this automatically degrades to creating a
109 normal directory. Among others /var/lib/machines is now
110 created like this at boot, should it be missing.
112 * The directory /var/lib/containers/ has been deprecated and
113 been replaced by /var/lib/machines. The term "machines" has
114 been used in the systemd context as generic term for both
115 VMs and containers, and hence appears more appropriate for
116 this, as the directory can also contain raw images bootable
119 * systemd-nspawn when invoked with -M but without --directory=
120 or --image= is now capable of searching for the container
121 root directory, subvolume or disk image automatically, in
122 /var/lib/machines. systemd-nspawn@.service has been updated
123 to make use of this, thus allowing it to be used for raw
126 * A new machines.target unit has been introduced that is
127 supposed to group all containers/VMs invoked as services on
128 the system. systemd-nspawn@.service has been updated to
131 * machinectl gained a new "start" command, for invoking a
132 container as a service. "machinectl start foo" is mostly
133 equivalent to "systemctl start systemd-nspawn@foo.service",
134 but handles escaping in a nicer way.
136 * systemd-nspawn will now mount most of the cgroupfs tree
137 read-only into each container, with the exception of the
138 container's own subtree in the name=systemd hierarchy.
140 * journald now sets the special FS_NOCOW file flag for its
141 journal files. This should improve performance on btrfs, by
142 avoiding heavy fragmentation when journald's write-pattern
143 is used on COW file systems. It degrades btrfs' data
144 integrity guarantees for the files to the same levels as for
145 ext3/ext4 however. This should be OK though as journald does
146 its own data integrity checks and all its objects are
147 checksummed on disk. Also, journald should handle btrfs disk
148 full events a lot more gracefully now, by processing SIGBUS
149 errors, and not relying on fallocate() anymore.
151 * When journald detects that journal files it is writing to
152 have been deleted it will immediately start new journal
155 * systemd now provides a way to store file descriptors
156 per-service in PID 1.This is useful for daemons to ensure
157 that fds they require are not lost during a daemon
158 restart. The fds are passed to the deamon on the next
159 invocation in the same way socket activation fds are
160 passed. This is now used by journald to ensure that the
161 various sockets connected to all the system's stdout/stderr
162 are not lost when journald is restarted. File descriptors
163 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
164 an extension to sd_notify(). Note that a limit is enforced
165 on the number of fds a service can store in PID 1, and it
166 defaults to 0, so that no fds may be stored, unless this is
167 explicitly turned on.
169 * The default TERM variable to use for units connected to a
170 terminal, when no other value is explicitly is set is now
171 vt220 rather than vt102. This should be fairly safe still,
172 but allows PgUp/PgDn work.
174 * The /etc/crypttab option header= as known from Debian is now
177 * "loginctl user-status" and "loginctl session-status" will
178 now show the last 10 lines of log messages of the
179 user/session following the status output. Similar,
180 "machinectl status" will show the last 10 log lines
181 associated with a virtual machine or container
182 service. (Note that this is usually not the log messages
183 done in the VM/container itself, but simply what the
184 container manager logs. For nspawn this includes all console
187 * "loginctl session-status" without further argument will now
188 show the status of the session of the caller. Similar,
189 "lock-session", "unlock-session", "activate",
190 "enable-linger", "disable-linger" may now be called without
191 session/user parameter in which case they apply to the
192 caller's session/user.
194 * An X11 session scriptlet is now shipped that uploads
195 $DISPLAY and $XAUTHORITY into the environment of the systemd
196 --user daemon if a session begins. This should improve
197 compatibility with X11 enabled applications run as systemd
200 * Generators are now subject to masking via /etc and /run, the
201 same way as unit files.
203 * networkd .network files gained support for configuring
204 per-link IPv4/IPv6 packet forwarding as well as IPv4
205 masquerading. This is by default turned on for veth links to
206 containers, as registered by systemd-nspawn. This means that
207 nspawn containers run with --network-veth will now get
208 automatic routed access to the host's networks without any
209 further configuration or setup, as long as networkd runs on
212 * systemd-nspawn gained the --port= (-p) switch to expose TCP
213 or UDP posts of a container on the host. With this in place
214 it is possible to run containers with private veth links
215 (--network-veth), and have their functionality exposed on
216 the host as if their services were running directly on the
219 * systemd-nspawn's --network-veth switch now gained a short
220 version "-n", since with the changes above it is now truly
221 useful out-of-the-box. The systemd-nspawn@.service has been
222 updated to make use of it too by default.
224 * systemd-nspawn will now maintain a per-image R/W lock, to
225 ensure that the same image is not started more than once
226 writable. (It's OK to run an image multiple times
227 simultaneously in read-only mode.)
229 * systemd-nspawn's --image= option is now capable of
230 dissecting and booting MBR and GPT disk images that contain
231 only a single active Linux partition. Previously it
232 supported only GPT disk images with proper GPT type
233 IDs. This allows running cloud images from major
234 distributions directly with systemd-nspawn, without
237 * In addition to collecting mouse dpi data in the udev
238 hardware database, there's now support for collecting angle
239 information for mouse scroll wheels. The database is
240 supposed to guarantee similar scrolling behavior on mice
241 that it knows about. There's also support for collecting
242 information about Touchpad types.
244 * udev's input_id built-in will now also collect touch screen
245 dimension data and attach it to probed devices.
247 * /etc/os-release gained support for a Distribution Privacy
250 * networkd gained support for creating "ipvlan", "gretap",
251 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
253 * systemd-tmpfiles gained support for "a" lines for setting
256 * systemd-nspawn will now mount /tmp in the container to
257 tmpfs, automatically.
259 * systemd now exposes the memory.usage_in_bytes cgroup
260 attribute and shows it for each service in the "systemctl
261 status" output, if available.
263 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
264 immediate reboot is triggered. This useful if shutdown is
265 hung and is unable to complete, to expedite the
266 operation. Note that this kind of reboot will still unmount
267 all file systems, and hence should not result in fsck being
270 * A .device unit for an optical block device will now be
271 considered active only when a medium is in the drive. Also,
272 mount units are now bound to their backing devices thus
273 triggering automatic unmounting when devices become
274 unavailable. With this in place systemd will now
275 automatically unmount left-over mounts when a CD-ROM is
276 ejected or an USB stick is yanked from the system.
278 * networkd-wait-online now has support for waiting for
279 specific interfaces only (with globbing), and for giving up
280 after a configurable timeout.
282 * networkd now exits when idle. It will be automatically
283 restarted as soon as interfaces show up, are removed or
284 change state. networkd will stay around as long as there is
285 at least one DHCP state machine or similar around, that keep
288 * networkd may now configure IPv6 link-local addressing in
289 addition to IPv4 link-local addressing.
291 * The IPv6 "token" for use in SLAAC may now be configured for
292 each .network interface in networkd.
294 * Routes configured with networkd may now be assigned a scope
297 * networkd's [Match] sections now support globbing and lists
298 of multiple space-separated matches per item.
300 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
301 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
302 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
303 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
304 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
305 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
306 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
307 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
308 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
309 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
310 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
311 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
312 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
313 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
314 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
315 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
316 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
317 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
318 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
319 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
320 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
321 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
322 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
323 Hoffmann, Zbigniew Jędrzejewski-Szmek
325 -- Berlin, 2015-02-16
329 * When querying unit file enablement status (for example via
330 "systemctl is-enabled"), a new state "indirect" is now known
331 which indicates that a unit might not be enabled itself, but
332 another unit listed in its Alias= setting might be.
334 * Similar to the various existing ConditionXYZ= settings for
335 units there are now matching AssertXYZ= settings. While
336 failing conditions cause a unit to be skipped, but its job
337 to succeed, failing assertions declared like this will cause
338 a unit start operation and its job to fail.
340 * hostnamed now knows a new chassis type "embedded".
342 * systemctl gained a new "edit" command. When used on a unit
343 file this allows extending unit files with .d/ drop-in
344 configuration snippets or editing the full file (after
345 copying it from /usr/lib to /etc). This will invoke the
346 user's editor (as configured with $EDITOR), and reload the
347 modified configuration after editing.
349 * "systemctl status" now shows the suggested enablement state
350 for a unit, as declared in the (usually vendor-supplied)
353 * nss-myhostname will now resolve the single-label host name
354 "gateway" to the locally configured default IP routing
355 gateways, ordered by their metrics. This assigns a stable
356 name to the used gateways, regardless which ones are
357 currently configured. Note that the name will only be
358 resolved after all other name sources (if nss-myhostname is
359 configured properly) and should hence not negatively impact
360 systems that use the single-label host name "gateway" in
363 * systemd-inhibit now allows filtering by mode when listing
366 * Scope and service units gained a new "Delegate" boolean
367 property, which when set allows processes running inside the
368 unit to further partition resources. This is primarily
369 useful for systemd user instances as well as container
372 * journald will now pick up audit messages directly from
373 the kernel, and log them like any other log message. The
374 audit fields are split up and fully indexed. This means that
375 journalctl in many ways is now a (nicer!) alternative to
376 ausearch, the traditional audit client. Note that this
377 implements only a minimal audit client, if you want the
378 special audit modes like reboot-on-log-overflow, please use
379 the traditional auditd instead, which can be used in
380 parallel to journald.
382 * The ConditionSecurity= unit file option now understands the
383 special string "audit" to check whether auditing is
386 * journalctl gained two new commands --vacuum-size= and
387 --vacuum-time= to delete old journal files until the
388 remaining ones take up no more the specified size on disk,
389 or are not older than the specified time.
391 * A new, native PPPoE library has been added to sd-network,
392 systemd's library of light-weight networking protocols. This
393 library will be used in a future version of networkd to
394 enable PPPoE communication without an external pppd daemon.
396 * The busctl tool now understands a new "capture" verb that
397 works similar to "monitor", but writes a packet capture
398 trace to STDOUT that can be redirected to a file which is
399 compatible with libcap's capture file format. This can then
400 be loaded in Wireshark and similar tools to inspect bus
403 * The busctl tool now understands a new "tree" verb that shows
404 the object trees of a specific service on the bus, or of all
407 * The busctl tool now understands a new "introspect" verb that
408 shows all interfaces and members of objects on the bus,
409 including their signature and values. This is particularly
410 useful to get more information about bus objects shown by
411 the new "busctl tree" command.
413 * The busctl tool now understands new verbs "call",
414 "set-property" and "get-property" for invoking bus method
415 calls, setting and getting bus object properties in a
418 * busctl gained a new --augment-creds= argument that controls
419 whether the tool shall augment credential information it
420 gets from the bus with data from /proc, in a possibly
423 * nspawn's --link-journal= switch gained two new values
424 "try-guest" and "try-host" that work like "guest" and
425 "host", but do not fail if the host has no persistent
426 journalling enabled. -j is now equivalent to
427 --link-journal=try-guest.
429 * macvlan network devices created by nspawn will now have
430 stable MAC addresses.
432 * A new SmackProcessLabel= unit setting has been added, which
433 controls the SMACK security label processes forked off by
434 the respective unit shall use.
436 * If compiled with --enable-xkbcommon, systemd-localed will
437 verify x11 keymap settings by compiling the given keymap. It
438 will spew out warnings if the compilation fails. This
439 requires libxkbcommon to be installed.
441 * When a coredump is collected a larger number of metadata
442 fields is now collected and included in the journal records
443 created for it. More specifically control group membership,
444 environment variables, memory maps, working directory,
445 chroot directory, /proc/$PID/status, and a list of open file
446 descriptors is now stored in the log entry.
448 * The udev hwdb now contains DPI information for mice. For
451 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
453 * All systemd programs that read standalone configuration
454 files in /etc now also support a corresponding series of
455 .conf.d configuration directories in /etc/, /run/,
456 /usr/local/lib/, /usr/lib/, and (if configured with
457 --enable-split-usr) /lib/. In particular, the following
458 configuration files now have corresponding configuration
459 directories: system.conf user.conf, logind.conf,
460 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
461 resolved.conf, timesyncd.conf, journal-remote.conf, and
462 journal-upload.conf. Note that distributions should use the
463 configuration directories in /usr/lib/; the directories in
464 /etc/ are reserved for the system administrator.
466 * systemd-rfkill will no longer take the rfkill device name
467 into account when storing rfkill state on disk, as the name
468 might be dynamically assigned and not stable. Instead, the
469 ID_PATH udev variable combined with the rfkill type (wlan,
470 bluetooth, ...) is used.
472 * A new service systemd-machine-id-commit.service has been
473 added. When used on systems where /etc is read-only during
474 boot, and /etc/machine-id is not initialized (but an empty
475 file), this service will copy the temporary machine ID
476 created as replacement into /etc after the system is fully
477 booted up. This is useful for systems that are freshly
478 installed with a non-initialized machine ID, but should get
479 a fixed machine ID for subsequent boots.
481 * networkd's .netdev files now provide a large set of
482 configuration parameters for VXLAN devices. Similar, the
483 bridge port cost parameter is now configurable in .network
484 files. There's also new support for configuring IP source
485 routing. networkd .link files gained support for a new
486 OriginalName= match that is useful to match against the
487 original interface name the kernel assigned. .network files
488 may include MTU= and MACAddress= fields for altering the MTU
489 and MAC address while being connected to a specific network
492 * The LUKS logic gained supported for configuring
493 UUID-specific key files. There's also new support for naming
494 LUKS device from the kernel command line, using the new
497 * Timer units may now be transiently created via the bus API
498 (this was previously already available for scope and service
499 units). In addition it is now possible to create multiple
500 transient units at the same time with a single bus call. The
501 "systemd-run" tool has been updated to make use of this for
502 running commands on a specified time, in at(1)-style.
504 * tmpfiles gained support for "t" lines, for assigning
505 extended attributes to files. Among other uses this may be
506 used to assign SMACK labels to files.
508 Contributions from: Alin Rauta, Alison Chaiken, Andrej
509 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
510 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
511 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
512 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
513 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
514 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
515 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
516 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
517 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
518 Michael Biebl, Michael Chapman, Michael Marineau, Michal
519 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
520 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
521 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
522 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
523 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
524 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
525 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
527 -- Berlin, 2014-12-10
531 * journalctl gained the new options -t/--identifier= to match
532 on the syslog identifier (aka "tag"), as well as --utc to
533 show log timestamps in the UTC timezone. journalctl now also
534 accepts -n/--lines=all to disable line capping in a pager.
536 * journalctl gained a new switch, --flush, that synchronously
537 flushes logs from /run/log/journal to /var/log/journal if
538 persistent storage is enabled. systemd-journal-flush.service
539 now waits until the operation is complete.
541 * Services can notify the manager before they start a reload
542 (by sending RELOADING=1) or shutdown (by sending
543 STOPPING=1). This allows the manager to track and show the
544 internal state of daemons and closes a race condition when
545 the process is still running but has closed its D-Bus
548 * Services with Type=oneshot do not have to have any ExecStart
551 * User units are now loaded also from
552 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
553 /run/systemd/user directory that was already previously
554 supported, but is under the control of the user.
556 * Job timeouts (i.e. time-outs on the time a job that is
557 queued stays in the run queue) can now optionally result in
558 immediate reboot or power-off actions (JobTimeoutAction= and
559 JobTimeoutRebootArgument=). This is useful on ".target"
560 units, to limit the maximum time a target remains
561 undispatched in the run queue, and to trigger an emergency
562 operation in such a case. This is now used by default to
563 turn off the system if boot-up (as defined by everything in
564 basic.target) hangs and does not complete for at least
565 15min. Also, if power-off or reboot hang for at least 30min
566 an immediate power-off/reboot operation is triggered. This
567 functionality is particularly useful to increase reliability
568 on embedded devices, but also on laptops which might
569 accidentally get powered on when carried in a backpack and
570 whose boot stays stuck in a hard disk encryption passphrase
573 * systemd-logind can be configured to also handle lid switch
574 events even when the machine is docked or multiple displays
575 are attached (HandleLidSwitchDocked= option).
577 * A helper binary and a service have been added which can be
578 used to resume from hibernation in the initramfs. A
579 generator will parse the resume= option on the kernel
580 command line to trigger resume.
582 * A user console daemon systemd-consoled has been
583 added. Currently, it is a preview, and will so far open a
584 single terminal on each session of the user marked as
585 Desktop=systemd-console.
587 * Route metrics can be specified for DHCP routes added by
590 * The SELinux context of socket-activated services can be set
591 from the information provided by the networking stack
592 (SELinuxContextFromNet= option).
594 * Userspace firmware loading support has been removed and
595 the minimum supported kernel version is thus bumped to 3.7.
597 * Timeout for udev workers has been increased from 1 to 3
598 minutes, but a warning will be printed after 1 minute to
599 help diagnose kernel modules that take a long time to load.
601 * Udev rules can now remove tags on devices with TAG-="foobar".
603 * systemd's readahead implementation has been removed. In many
604 circumstances it didn't give expected benefits even for
605 rotational disk drives and was becoming less relevant in the
606 age of SSDs. As none of the developers has been using
607 rotating media anymore, and nobody stepped up to actively
608 maintain this component of systemd it has now been removed.
610 * Swap units can use Options= to specify discard options.
611 Discard options specified for swaps in /etc/fstab are now
614 * Docker containers are now detected as a separate type of
617 * The Password Agent protocol gained support for queries where
618 the user input is shown, useful e.g. for user names.
619 systemd-ask-password gained a new --echo option to turn that
622 * The default sysctl.d/ snippets will now set:
624 net.core.default_qdisc = fq_codel
626 This selects Fair Queuing Controlled Delay as the default
627 queuing discipline for network interfaces. fq_codel helps
628 fight the network bufferbloat problem. It is believed to be
629 a good default with no tuning required for most workloads.
630 Downstream distributions may override this choice. On 10Gbit
631 servers that do not do forwarding, "fq" may perform better.
632 Systems without a good clocksource should use "pfifo_fast".
634 * If kdbus is enabled during build a new option BusPolicy= is
635 available for service units, that allows locking all service
636 processes into a stricter bus policy, in order to limit
637 access to various bus services, or even hide most of them
638 from the service's view entirely.
640 * networkctl will now show the .network and .link file
641 networkd has applied to a specific interface.
643 * sd-login gained a new API call sd_session_get_desktop() to
644 query which desktop environment has been selected for a
647 * UNIX utmp support is now compile-time optional to support
650 * systemctl gained two new commands "add-wants" and
651 "add-requires" for pulling in units from specific targets
654 * If the word "rescue" is specified on the kernel command line
655 the system will now boot into rescue mode (aka
656 rescue.target), which was previously available only by
657 specifying "1" or "systemd.unit=rescue.target" on the kernel
658 command line. This new kernel command line option nicely
659 mirrors the already existing "emergency" kernel command line
662 * New kernel command line options mount.usr=, mount.usrflags=,
663 mount.usrfstype= have been added that match root=, rootflags=,
664 rootfstype= but allow mounting a specific file system to
667 * The $NOTIFY_SOCKET is now also passed to control processes of
668 services, not only the main process.
670 * This version reenables support for fsck's -l switch. This
671 means at least version v2.25 of util-linux is required for
672 operation, otherwise dead-locks on device nodes may
673 occur. Again: you need to update util-linux to at least
674 v2.25 when updating systemd to v217.
676 * The "multi-seat-x" tool has been removed from systemd, as
677 its functionality has been integrated into X servers 1.16,
678 and the tool is hence redundant. It is recommended to update
679 display managers invoking this tool to simply invoke X
680 directly from now on, again.
682 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
683 message flag has been added for all of systemd's PolicyKit
684 authenticated method calls has been added. In particular
685 this now allows optional interactive authorization via
686 PolicyKit for many of PID1's privileged operations such as
687 unit file enabling and disabling.
689 * "udevadm hwdb --update" learnt a new switch "--usr" for
690 placing the rebuilt hardware database in /usr instead of
691 /etc. When used only hardware database entries stored in
692 /usr will be used, and any user database entries in /etc are
693 ignored. This functionality is useful for vendors to ship a
694 pre-built database on systems where local configuration is
695 unnecessary or unlikely.
697 * Calendar time specifications in .timer units now also
698 understand the strings "semi-annually", "quarterly" and
699 "minutely" as shortcuts (in addition to the preexisting
700 "anually", "hourly", ...).
702 * systemd-tmpfiles will now correctly create files in /dev
703 at boot which are marked for creation only at boot. It is
704 recommended to always create static device nodes with 'c!'
705 and 'b!', so that they are created only at boot and not
706 overwritten at runtime.
708 * When the watchdog logic is used for a service (WatchdogSec=)
709 and the watchdog timeout is hit the service will now be
710 terminated with SIGABRT (instead of just SIGTERM), in order
711 to make sure a proper coredump and backtrace is
712 generated. This ensures that hanging services will result in
713 similar coredump/backtrace behaviour as services that hit a
716 Contributions from: Andreas Henriksson, Andrei Borzenkov,
717 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
718 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
719 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
720 Herrmann, David Sommerseth, David Strauss, Emil Renner
721 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
722 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
723 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
724 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
725 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
726 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
727 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
728 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
729 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
730 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
731 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
732 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
733 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
734 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
735 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
736 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
739 -- Berlin, 2014-10-28
743 * timedated no longer reads NTP implementation unit names from
744 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
745 implementations should add a
747 Conflicts=systemd-timesyncd.service
749 to their unit files to take over and replace systemd's NTP
750 default functionality.
752 * systemd-sysusers gained a new line type "r" for configuring
753 which UID/GID ranges to allocate system users/groups
754 from. Lines of type "u" may now add an additional column
755 that specifies the home directory for the system user to be
756 created. Also, systemd-sysusers may now optionally read user
757 information from STDIN instead of a file. This is useful for
758 invoking it from RPM preinst scriptlets that need to create
759 users before the first RPM file is installed since these
760 files might need to be owned by them. A new
761 %sysusers_create_inline RPM macro has been introduced to do
762 just that. systemd-sysusers now updates the shadow files as
763 well as the user/group databases, which should enhance
764 compatibility with certain tools like grpck.
766 * A number of bus APIs of PID 1 now optionally consult
767 PolicyKit to permit access for otherwise unprivileged
768 clients under certain conditions. Note that this currently
769 doesn't support interactive authentication yet, but this is
770 expected to be added eventually, too.
772 * /etc/machine-info now has new fields for configuring the
773 deployment environment of the machine, as well as the
774 location of the machine. hostnamectl has been updated with
775 new command to update these fields.
777 * systemd-timesyncd has been updated to automatically acquire
778 NTP server information from systemd-networkd, which might
779 have been discovered via DHCP.
781 * systemd-resolved now includes a caching DNS stub resolver
782 and a complete LLMNR name resolution implementation. A new
783 NSS module "nss-resolve" has been added which can be used
784 instead of glibc's own "nss-dns" to resolve hostnames via
785 systemd-resolved. Hostnames, addresses and arbitrary RRs may
786 be resolved via systemd-resolved D-Bus APIs. In contrast to
787 the glibc internal resolver systemd-resolved is aware of
788 multi-homed system, and keeps DNS server and caches separate
789 and per-interface. Queries are sent simultaneously on all
790 interfaces that have DNS servers configured, in order to
791 properly handle VPNs and local LANs which might resolve
792 separate sets of domain names. systemd-resolved may acquire
793 DNS server information from systemd-networkd automatically,
794 which in turn might have discovered them via DHCP. A tool
795 "systemd-resolve-host" has been added that may be used to
796 query the DNS logic in resolved. systemd-resolved implements
797 IDNA and automatically uses IDNA or UTF-8 encoding depending
798 on whether classic DNS or LLMNR is used as transport. In the
799 next releases we intend to add a DNSSEC and mDNS/DNS-SD
800 implementation to systemd-resolved.
802 * A new NSS module nss-mymachines has been added, that
803 automatically resolves the names of all local registered
804 containers to their respective IP addresses.
806 * A new client tool "networkctl" for systemd-networkd has been
807 added. It currently is entirely passive and will query
808 networking configuration from udev, rtnetlink and networkd,
809 and present it to the user in a very friendly
810 way. Eventually, we hope to extend it to become a full
811 control utility for networkd.
813 * .socket units gained a new DeferAcceptSec= setting that
814 controls the kernels' TCP_DEFER_ACCEPT sockopt for
815 TCP. Similar, support for controlling TCP keep-alive
816 settings has been added (KeepAliveTimeSec=,
817 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
818 turning off Nagle's algorithm on TCP has been added
821 * logind learned a new session type "web", for use in projects
822 like Cockpit which register web clients as PAM sessions.
824 * timer units with at least one OnCalendar= setting will now
825 be started only after timer-sync.target has been
826 reached. This way they will not elapse before the system
827 clock has been corrected by a local NTP client or
828 similar. This is particular useful on RTC-less embedded
829 machines, that come up with an invalid system clock.
831 * systemd-nspawn's --network-veth= switch should now result in
832 stable MAC addresses for both the outer and the inner side
835 * systemd-nspawn gained a new --volatile= switch for running
836 container instances with /etc or /var unpopulated.
838 * The kdbus client code has been updated to use the new Linux
839 3.17 memfd subsystem instead of the old kdbus-specific one.
841 * systemd-networkd's DHCP client and server now support
842 FORCERENEW. There are also new configuration options to
843 configure the vendor client identifier and broadcast mode
846 * systemd will no longer inform the kernel about the current
847 timezone, as this is necessarily incorrect and racy as the
848 kernel has no understanding of DST and similar
849 concepts. This hence means FAT timestamps will be always
850 considered UTC, similar to what Android is already
851 doing. Also, when the RTC is configured to the local time
852 (rather than UTC) systemd will never synchronize back to it,
853 as this might confuse Windows at a later boot.
855 * systemd-analyze gained a new command "verify" for offline
856 validation of unit files.
858 * systemd-networkd gained support for a couple of additional
859 settings for bonding networking setups. Also, the metric for
860 statically configured routes may now be configured. For
861 network interfaces where this is appropriate the peer IP
862 address may now be configured.
864 * systemd-networkd's DHCP client will no longer request
865 broadcasting by default, as this tripped up some networks.
866 For hardware where broadcast is required the feature should
867 be switched back on using RequestBroadcast=yes.
869 * systemd-networkd will now set up IPv4LL addresses (when
870 enabled) even if DHCP is configured successfully.
872 * udev will now default to respect network device names given
873 by the kernel when the kernel indicates that these are
874 predictable. This behavior can be tweaked by changing
875 NamePolicy= in the relevant .link file.
877 * A new library systemd-terminal has been added that
878 implements full TTY stream parsing and rendering. This
879 library is supposed to be used later on for implementing a
880 full userspace VT subsystem, replacing the current kernel
883 * A new tool systemd-journal-upload has been added to push
884 journal data to a remote system running
885 systemd-journal-remote.
887 * journald will no longer forward all local data to another
888 running syslog daemon. This change has been made because
889 rsyslog (which appears to be the most commonly used syslog
890 implementation these days) no longer makes use of this, and
891 instead pulls the data out of the journal on its own. Since
892 forwarding the messages to a non-existent syslog server is
893 more expensive than we assumed we have now turned this
894 off. If you run a syslog server that is not a recent rsyslog
895 version, you have to turn this option on again
896 (ForwardToSyslog= in journald.conf).
898 * journald now optionally supports the LZ4 compressor for
899 larger journal fields. This compressor should perform much
900 better than XZ which was the previous default.
902 * machinectl now shows the IP addresses of local containers,
903 if it knows them, plus the interface name of the container.
905 * A new tool "systemd-escape" has been added that makes it
906 easy to escape strings to build unit names and similar.
908 * sd_notify() messages may now include a new ERRNO= field
909 which is parsed and collected by systemd and shown among the
910 "systemctl status" output for a service.
912 * A new component "systemd-firstboot" has been added that
913 queries the most basic systemd information (timezone,
914 hostname, root password) interactively on first
915 boot. Alternatively it may also be used to provision these
916 things offline on OS images installed into directories.
918 * The default sysctl.d/ snippets will now set
920 net.ipv4.conf.default.promote_secondaries=1
922 This has the benefit of no flushing secondary IP addresses
923 when primary addresses are removed.
925 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
926 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
927 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
928 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
929 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
930 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
931 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
932 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
933 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
934 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
935 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
936 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
937 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
938 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
939 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
941 -- Berlin, 2014-08-19
945 * A new tool systemd-sysusers has been added. This tool
946 creates system users and groups in /etc/passwd and
947 /etc/group, based on static declarative system user/group
948 definitions in /usr/lib/sysusers.d/. This is useful to
949 enable factory resets and volatile systems that boot up with
950 an empty /etc directory, and thus need system users and
951 groups created during early boot. systemd now also ships
952 with two default sysusers.d/ files for the most basic
953 users and groups systemd and the core operating system
956 * A new tmpfiles snippet has been added that rebuilds the
957 essential files in /etc on boot, should they be missing.
959 * A directive for ensuring automatic clean-up of
960 /var/cache/man/ has been removed from the default
961 configuration. This line should now be shipped by the man
962 implementation. The necessary change has been made to the
963 man-db implementation. Note that you need to update your man
964 implementation to one that ships this line, otherwise no
965 automatic clean-up of /var/cache/man will take place.
967 * A new condition ConditionNeedsUpdate= has been added that
968 may conditionalize services to only run when /etc or /var
969 are "older" than the vendor operating system resources in
970 /usr. This is useful for reconstructing or updating /etc
971 after an offline update of /usr or a factory reset, on the
972 next reboot. Services that want to run once after such an
973 update or reset should use this condition and order
974 themselves before the new systemd-update-done.service, which
975 will mark the two directories as fully updated. A number of
976 service files have been added making use of this, to rebuild
977 the udev hardware database, the journald message catalog and
978 dynamic loader cache (ldconfig). The systemd-sysusers tool
979 described above also makes use of this now. With this in
980 place it is now possible to start up a minimal operating
981 system with /etc empty cleanly. For more information on the
982 concepts involved see this recent blog story:
984 http://0pointer.de/blog/projects/stateless.html
986 * A new system group "input" has been introduced, and all
987 input device nodes get this group assigned. This is useful
988 for system-level software to get access to input devices. It
989 complements what is already done for "audio" and "video".
991 * systemd-networkd learnt minimal DHCPv4 server support in
992 addition to the existing DHCPv4 client support. It also
993 learnt DHCPv6 client and IPv6 Router Solicitation client
994 support. The DHCPv4 client gained support for static routes
995 passed in from the server. Note that the [DHCPv4] section
996 known in older systemd-networkd versions has been renamed to
997 [DHCP] and is now also used by the DHCPv6 client. Existing
998 .network files using settings of this section should be
999 updated, though compatibility is maintained. Optionally, the
1000 client hostname may now be sent to the DHCP server.
1002 * networkd gained support for vxlan virtual networks as well
1003 as tun/tap and dummy devices.
1005 * networkd gained support for automatic allocation of address
1006 ranges for interfaces from a system-wide pool of
1007 addresses. This is useful for dynamically managing a large
1008 number of interfaces with a single network configuration
1009 file. In particular this is useful to easily assign
1010 appropriate IP addresses to the veth links of a large number
1011 of nspawn instances.
1013 * RPM macros for processing sysusers, sysctl and binfmt
1014 drop-in snippets at package installation time have been
1017 * The /etc/os-release file should now be placed in
1018 /usr/lib/os-release. The old location is automatically
1019 created as symlink. /usr/lib is the more appropriate
1020 location of this file, since it shall actually describe the
1021 vendor operating system shipped in /usr, and not the
1022 configuration stored in /etc.
1024 * .mount units gained a new boolean SloppyOptions= setting
1025 that maps to mount(8)'s -s option which enables permissive
1026 parsing of unknown mount options.
1028 * tmpfiles learnt a new "L+" directive which creates a symlink
1029 but (unlike "L") deletes a pre-existing file first, should
1030 it already exist and not already be the correct
1031 symlink. Similar, "b+", "c+" and "p+" directives have been
1032 added as well, which create block and character devices, as
1033 well as fifos in the filesystem, possibly removing any
1034 pre-existing files of different types.
1036 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
1037 'argument' field (which so far specified the source to
1038 symlink/copy the files from) is now optional. If omitted the
1039 same file os copied from /usr/share/factory/ suffixed by the
1040 full destination path. This is useful for populating /etc
1041 with essential files, by copying them from vendor defaults
1042 shipped in /usr/share/factory/etc.
1044 * A new command "systemctl preset-all" has been added that
1045 applies the service preset settings to all installed unit
1046 files. A new switch --preset-mode= has been added that
1047 controls whether only enable or only disable operations
1050 * A new command "systemctl is-system-running" has been added
1051 that allows checking the overall state of the system, for
1052 example whether it is fully up and running.
1054 * When the system boots up with an empty /etc, the equivalent
1055 to "systemctl preset-all" is executed during early boot, to
1056 make sure all default services are enabled after a factory
1059 * systemd now contains a minimal preset file that enables the
1060 most basic services systemd ships by default.
1062 * Unit files' [Install] section gained a new DefaultInstance=
1063 field for defining the default instance to create if a
1064 template unit is enabled with no instance specified.
1066 * A new passive target cryptsetup-pre.target has been added
1067 that may be used by services that need to make they run and
1068 finish before the first LUKS cryptographic device is set up.
1070 * The /dev/loop-control and /dev/btrfs-control device nodes
1071 are now owned by the "disk" group by default, opening up
1072 access to this group.
1074 * systemd-coredump will now automatically generate a
1075 stack trace of all core dumps taking place on the system,
1076 based on elfutils' libdw library. This stack trace is logged
1079 * systemd-coredump may now optionally store coredumps directly
1080 on disk (in /var/lib/systemd/coredump, possibly compressed),
1081 instead of storing them unconditionally in the journal. This
1082 mode is the new default. A new configuration file
1083 /etc/systemd/coredump.conf has been added to configure this
1084 and other parameters of systemd-coredump.
1086 * coredumpctl gained a new "info" verb to show details about a
1087 specific coredump. A new switch "-1" has also been added
1088 that makes sure to only show information about the most
1089 recent entry instead of all entries. Also, as the tool is
1090 generally useful now the "systemd-" prefix of the binary
1091 name has been removed. Distributions that want to maintain
1092 compatibility with the old name should add a symlink from
1093 the old name to the new name.
1095 * journald's SplitMode= now defaults to "uid". This makes sure
1096 that unprivileged users can access their own coredumps with
1097 coredumpctl without restrictions.
1099 * New kernel command line options "systemd.wants=" (for
1100 pulling an additional unit during boot), "systemd.mask="
1101 (for masking a specific unit for the boot), and
1102 "systemd.debug-shell" (for enabling the debug shell on tty9)
1103 have been added. This is implemented in the new generator
1104 "systemd-debug-generator".
1106 * systemd-nspawn will now by default filter a couple of
1107 syscalls for containers, among them those required for
1108 kernel module loading, direct x86 IO port access, swap
1109 management, and kexec. Most importantly though
1110 open_by_handle_at() is now prohibited for containers,
1111 closing a hole similar to a recently discussed vulnerability
1112 in docker regarding access to files on file hierarchies the
1113 container should normally not have access to. Note that for
1114 nspawn we generally make no security claims anyway (and
1115 this is explicitly documented in the man page), so this is
1116 just a fix for one of the most obvious problems.
1118 * A new man page file-hierarchy(7) has been added that
1119 contains a minimized, modernized version of the file system
1120 layout systemd expects, similar in style to the FHS
1121 specification or hier(5). A new tool systemd-path(1) has
1122 been added to query many of these paths for the local
1125 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
1126 longer done. Since the directory now has a per-user size
1127 limit, and is cleaned on logout this appears unnecessary,
1128 in particular since this now brings the lifecycle of this
1129 directory closer in line with how IPC objects are handled.
1131 * systemd.pc now exports a number of additional directories,
1132 including $libdir (which is useful to identify the library
1133 path for the primary architecture of the system), and a
1134 couple of drop-in directories.
1136 * udev's predictable network interface names now use the dev_port
1137 sysfs attribute, introduced in linux 3.15 instead of dev_id to
1138 distinguish between ports of the same PCI function. dev_id should
1139 only be used for ports using the same HW address, hence the need
1142 * machined has been updated to export the OS version of a
1143 container (read from /etc/os-release and
1144 /usr/lib/os-release) on the bus. This is now shown in
1145 "machinectl status" for a machine.
1147 * A new service setting RestartForceExitStatus= has been
1148 added. If configured to a set of exit signals or process
1149 return values, the service will be restarted when the main
1150 daemon process exits with any of them, regardless of the
1153 * systemctl's -H switch for connecting to remote systemd
1154 machines has been extended so that it may be used to
1155 directly connect to a specific container on the
1156 host. "systemctl -H root@foobar:waldi" will now connect as
1157 user "root" to host "foobar", and then proceed directly to
1158 the container named "waldi". Note that currently you have to
1159 authenticate as user "root" for this to work, as entering
1160 containers is a privileged operation.
1162 Contributions from: Andreas Henriksson, Benjamin Steinwender,
1163 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
1164 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
1165 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
1166 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
1167 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
1168 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
1169 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
1170 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
1171 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
1172 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
1173 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
1175 -- Berlin, 2014-07-03
1179 * As an experimental feature, udev now tries to lock the
1180 disk device node (flock(LOCK_SH|LOCK_NB)) while it
1181 executes events for the disk or any of its partitions.
1182 Applications like partitioning programs can lock the
1183 disk device node (flock(LOCK_EX)) and claim temporary
1184 device ownership that way; udev will entirely skip all event
1185 handling for this disk and its partitions. If the disk
1186 was opened for writing, the close will trigger a partition
1187 table rescan in udev's "watch" facility, and if needed
1188 synthesize "change" events for the disk and all its partitions.
1189 This is now unconditionally enabled, and if it turns out to
1190 cause major problems, we might turn it on only for specific
1191 devices, or might need to disable it entirely. Device Mapper
1192 devices are excluded from this logic.
1194 * We temporarily dropped the "-l" switch for fsck invocations,
1195 since they collide with the flock() logic above. util-linux
1196 upstream has been changed already to avoid this conflict,
1197 and we will readd "-l" as soon as util-linux with this
1198 change has been released.
1200 * The dependency on libattr has been removed. Since a long
1201 time, the extended attribute calls have moved to glibc, and
1202 libattr is thus unnecessary.
1204 * Virtualization detection works without priviliges now. This
1205 means the systemd-detect-virt binary no longer requires
1206 CAP_SYS_PTRACE file capabilities, and our daemons can run
1207 with fewer privileges.
1209 * systemd-networkd now runs under its own "systemd-network"
1210 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
1211 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
1212 loses the ability to write to files owned by root this way.
1214 * Similar, systemd-resolved now runs under its own
1215 "systemd-resolve" user with no capabilities remaining.
1217 * Similar, systemd-bus-proxyd now runs under its own
1218 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
1220 * systemd-networkd gained support for setting up "veth"
1221 virtual ethernet devices for container connectivity, as well
1222 as GRE and VTI tunnels.
1224 * systemd-networkd will no longer automatically attempt to
1225 manually load kernel modules necessary for certain tunnel
1226 transports. Instead, it is assumed the kernel loads them
1227 automatically when required. This only works correctly on
1228 very new kernels. On older kernels, please consider adding
1229 the kernel modules to /etc/modules-load.d/ as a work-around.
1231 * The resolv.conf file systemd-resolved generates has been
1232 moved to /run/systemd/resolve/. If you have a symlink from
1233 /etc/resolv.conf, it might be necessary to correct it.
1235 * Two new service settings, ProtectHome= and ProtectSystem=,
1236 have been added. When enabled, they will make the user data
1237 (such as /home) inaccessible or read-only and the system
1238 (such as /usr) read-only, for specific services. This allows
1239 very light-weight per-service sandboxing to avoid
1240 modifications of user data or system files from
1241 services. These two new switches have been enabled for all
1242 of systemd's long-running services, where appropriate.
1244 * Socket units gained new SocketUser= and SocketGroup=
1245 settings to set the owner user and group of AF_UNIX sockets
1246 and FIFOs in the file system.
1248 * Socket units gained a new RemoveOnStop= setting. If enabled,
1249 all FIFOS and sockets in the file system will be removed
1250 when the specific socket unit is stopped.
1252 * Socket units gained a new Symlinks= setting. It takes a list
1253 of symlinks to create to file system sockets or FIFOs
1254 created by the specific Unix sockets. This is useful to
1255 manage symlinks to socket nodes with the same life-cycle as
1258 * The /dev/log socket and /dev/initctl FIFO have been moved to
1259 /run, and have been replaced by symlinks. This allows
1260 connecting to these facilities even if PrivateDevices=yes is
1261 used for a service (which makes /dev/log itself unavailable,
1262 but /run is left). This also has the benefit of ensuring
1263 that /dev only contains device nodes, directories and
1264 symlinks, and nothing else.
1266 * sd-daemon gained two new calls sd_pid_notify() and
1267 sd_pid_notifyf(). They are similar to sd_notify() and
1268 sd_notifyf(), but allow overriding of the source PID of
1269 notification messages if permissions permit this. This is
1270 useful to send notify messages on behalf of a different
1271 process (for example, the parent process). The
1272 systemd-notify tool has been updated to make use of this
1273 when sending messages (so that notification messages now
1274 originate from the shell script invoking systemd-notify and
1275 not the systemd-notify process itself. This should minimize
1276 a race where systemd fails to associate notification
1277 messages to services when the originating process already
1280 * A new "on-abnormal" setting for Restart= has been added. If
1281 set, it will result in automatic restarts on all "abnormal"
1282 reasons for a process to exit, which includes unclean
1283 signals, core dumps, timeouts and watchdog timeouts, but
1284 does not include clean and unclean exit codes or clean
1285 signals. Restart=on-abnormal is an alternative for
1286 Restart=on-failure for services that shall be able to
1287 terminate and avoid restarts on certain errors, by
1288 indicating so with an unclean exit code. Restart=on-failure
1289 or Restart=on-abnormal is now the recommended setting for
1290 all long-running services.
1292 * If the InaccessibleDirectories= service setting points to a
1293 mount point (or if there are any submounts contained within
1294 it), it is now attempted to completely unmount it, to make
1295 the file systems truly unavailable for the respective
1298 * The ReadOnlyDirectories= service setting and
1299 systemd-nspawn's --read-only parameter are now recursively
1300 applied to all submounts, too.
1302 * Mount units may now be created transiently via the bus APIs.
1304 * The support for SysV and LSB init scripts has been removed
1305 from the systemd daemon itself. Instead, it is now
1306 implemented as a generator that creates native systemd units
1307 from these scripts when needed. This enables us to remove a
1308 substantial amount of legacy code from PID 1, following the
1309 fact that many distributions only ship a very small number
1310 of LSB/SysV init scripts nowadays.
1312 * Privileged Xen (dom0) domains are not considered
1313 virtualization anymore by the virtualization detection
1314 logic. After all, they generally have unrestricted access to
1315 the hardware and usually are used to manage the unprivileged
1318 * systemd-tmpfiles gained a new "C" line type, for copying
1319 files or entire directories.
1321 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
1322 lines. So far, they have been non-globbing versions of the
1323 latter, and have thus been redundant. In future, it is
1324 recommended to only use "z". "m" has hence been removed
1325 from the documentation, even though it stays supported.
1327 * A tmpfiles snippet to recreate the most basic structure in
1328 /var has been added. This is enough to create the /var/run →
1329 /run symlink and create a couple of structural
1330 directories. This allows systems to boot up with an empty or
1331 volatile /var. Of course, while with this change, the core OS
1332 now is capable with dealing with a volatile /var, not all
1333 user services are ready for it. However, we hope that sooner
1334 or later, many service daemons will be changed upstream so
1335 that they are able to automatically create their necessary
1336 directories in /var at boot, should they be missing. This is
1337 the first step to allow state-less systems that only require
1338 the vendor image for /usr to boot.
1340 * systemd-nspawn has gained a new --tmpfs= switch to mount an
1341 empty tmpfs instance to a specific directory. This is
1342 particularly useful for making use of the automatic
1343 reconstruction of /var (see above), by passing --tmpfs=/var.
1345 * Access modes specified in tmpfiles snippets may now be
1346 prefixed with "~", which indicates that they shall be masked
1347 by whether the existing file or directory is currently
1348 writable, readable or executable at all. Also, if specified,
1349 the sgid/suid/sticky bits will be masked for all
1352 * A new passive target unit "network-pre.target" has been
1353 added which is useful for services that shall run before any
1354 network is configured, for example firewall scripts.
1356 * The "floppy" group that previously owned the /dev/fd*
1357 devices is no longer used. The "disk" group is now used
1358 instead. Distributions should probably deprecate usage of
1361 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
1362 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
1363 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
1364 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
1365 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
1366 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
1369 -- Berlin, 2014-06-11
1373 * A new "systemd-timesyncd" daemon has been added for
1374 synchronizing the system clock across the network. It
1375 implements an SNTP client. In contrast to NTP
1376 implementations such as chrony or the NTP reference server,
1377 this only implements a client side, and does not bother with
1378 the full NTP complexity, focusing only on querying time from
1379 one remote server and synchronizing the local clock to
1380 it. Unless you intend to serve NTP to networked clients or
1381 want to connect to local hardware clocks, this simple NTP
1382 client should be more than appropriate for most
1383 installations. The daemon runs with minimal privileges, and
1384 has been hooked up with networkd to only operate when
1385 network connectivity is available. The daemon saves the
1386 current clock to disk every time a new NTP sync has been
1387 acquired, and uses this to possibly correct the system clock
1388 early at bootup, in order to accommodate for systems that
1389 lack an RTC such as the Raspberry Pi and embedded devices,
1390 and to make sure that time monotonically progresses on these
1391 systems, even if it is not always correct. To make use of
1392 this daemon, a new system user and group "systemd-timesync"
1393 needs to be created on installation of systemd.
1395 * The queue "seqnum" interface of libudev has been disabled, as
1396 it was generally incompatible with device namespacing as
1397 sequence numbers of devices go "missing" if the devices are
1398 part of a different namespace.
1400 * "systemctl list-timers" and "systemctl list-sockets" gained
1401 a --recursive switch for showing units of these types also
1402 for all local containers, similar in style to the already
1403 supported --recursive switch for "systemctl list-units".
1405 * A new RebootArgument= setting has been added for service
1406 units, which may be used to specify a kernel reboot argument
1407 to use when triggering reboots with StartLimitAction=.
1409 * A new FailureAction= setting has been added for service
1410 units which may be used to specify an operation to trigger
1411 when a service fails. This works similarly to
1412 StartLimitAction=, but unlike it, controls what is done
1413 immediately rather than only after several attempts to
1414 restart the service in question.
1416 * hostnamed got updated to also expose the kernel name,
1417 release, and version on the bus. This is useful for
1418 executing commands like hostnamectl with the -H switch.
1419 systemd-analyze makes use of this to properly display
1420 details when running non-locally.
1422 * The bootchart tool can now show cgroup information in the
1423 graphs it generates.
1425 * The CFS CPU quota cgroup attribute is now exposed for
1426 services. The new CPUQuota= switch has been added for this
1427 which takes a percentage value. Setting this will have the
1428 result that a service may never get more CPU time than the
1429 specified percentage, even if the machine is otherwise idle.
1431 * systemd-networkd learned IPIP and SIT tunnel support.
1433 * LSB init scripts exposing a dependency on $network will now
1434 get a dependency on network-online.target rather than simply
1435 network.target. This should bring LSB handling closer to
1436 what it was on SysV systems.
1438 * A new fsck.repair= kernel option has been added to control
1439 how fsck shall deal with unclean file systems at boot.
1441 * The (.ini) configuration file parser will now silently
1442 ignore sections whose name begins with "X-". This may be
1443 used to maintain application-specific extension sections in unit
1446 * machined gained a new API to query the IP addresses of
1447 registered containers. "machinectl status" has been updated
1448 to show these addresses in its output.
1450 * A new call sd_uid_get_display() has been added to the
1451 sd-login APIs for querying the "primary" session of a
1452 user. The "primary" session of the user is elected from the
1453 user's sessions and generally a graphical session is
1454 preferred over a text one.
1456 * A minimal systemd-resolved daemon has been added. It
1457 currently simply acts as a companion to systemd-networkd and
1458 manages resolv.conf based on per-interface DNS
1459 configuration, possibly supplied via DHCP. In the long run
1460 we hope to extend this into a local DNSSEC enabled DNS and
1463 * The systemd-networkd-wait-online tool is now enabled by
1464 default. It will delay network-online.target until a network
1465 connection has been configured. The tool primarily integrates
1466 with networkd, but will also make a best effort to make sense
1467 of network configuration performed in some other way.
1469 * Two new service options StartupCPUShares= and
1470 StartupBlockIOWeight= have been added that work similarly to
1471 CPUShares= and BlockIOWeight= however only apply during
1472 system startup. This is useful to prioritize certain services
1473 differently during bootup than during normal runtime.
1475 * hostnamed has been changed to prefer the statically
1476 configured hostname in /etc/hostname (unless set to
1477 'localhost' or empty) over any dynamic one supplied by
1478 dhcp. With this change, the rules for picking the hostname
1479 match more closely the rules of other configuration settings
1480 where the local administrator's configuration in /etc always
1481 overrides any other settings.
1483 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
1484 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
1485 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
1486 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
1487 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
1488 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
1489 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
1490 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
1491 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
1492 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
1493 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
1494 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
1495 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
1496 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
1497 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
1498 Lindskog, WaLyong Cho, Will Woods, Zbigniew
1501 -- Beijing, 2014-05-28
1505 * When restoring the screen brightness at boot, stay away from
1506 the darkest setting or from the lowest 5% of the available
1507 range, depending on which is the larger value of both. This
1508 should effectively protect the user from rebooting into a
1509 black screen, should the brightness have been set to minimum
1512 * sd-login gained a new sd_machine_get_class() call to
1513 determine the class ("vm" or "container") of a machine
1514 registered with machined.
1516 * sd-login gained new calls
1517 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
1518 to query the identity of the peer of a local AF_UNIX
1519 connection. They operate similarly to their sd_pid_get_xyz()
1522 * PID 1 will now maintain a system-wide system state engine
1523 with the states "starting", "running", "degraded",
1524 "maintenance", "stopping". These states are bound to system
1525 startup, normal runtime, runtime with at least one failed
1526 service, rescue/emergency mode and system shutdown. This
1527 state is shown in the "systemctl status" output when no unit
1528 name is passed. It is useful to determine system state, in
1529 particularly when doing so for many systems or containers at
1532 * A new command "list-machines" has been added to "systemctl"
1533 that lists all local OS containers and shows their system
1534 state (see above), if systemd runs inside of them.
1536 * systemctl gained a new "-r" switch to recursively enumerate
1537 units on all local containers, when used with the
1538 "list-unit" command (which is the default one that is
1539 executed when no parameters are specified).
1541 * The GPT automatic partition discovery logic will now honour
1542 two GPT partition flags: one may be set on a partition to
1543 cause it to be mounted read-only, and the other may be set
1544 on a partition to ignore it during automatic discovery.
1546 * Two new GPT type UUIDs have been added for automatic root
1547 partition discovery, for 32-bit and 64-bit ARM. This is not
1548 particularly useful for discovering the root directory on
1549 these architectures during bare-metal boots (since UEFI is
1550 not common there), but still very useful to allow booting of
1551 ARM disk images in nspawn with the -i option.
1553 * MAC addresses of interfaces created with nspawn's
1554 --network-interface= switch will now be generated from the
1555 machine name, and thus be stable between multiple invocations
1558 * logind will now automatically remove all IPC objects owned
1559 by a user if she or he fully logs out. This makes sure that
1560 users who are logged out cannot continue to consume IPC
1561 resources. This covers SysV memory, semaphores and message
1562 queues as well as POSIX shared memory and message
1563 queues. Traditionally, SysV and POSIX IPC had no life-cycle
1564 limits. With this functionality, that is corrected. This may
1565 be turned off by using the RemoveIPC= switch of logind.conf.
1567 * The systemd-machine-id-setup and tmpfiles tools gained a
1568 --root= switch to operate on a specific root directory,
1571 * journald can now forward logged messages to the TTYs of all
1572 logged in users ("wall"). This is the default for all
1573 emergency messages now.
1575 * A new tool systemd-journal-remote has been added to stream
1576 journal log messages across the network.
1578 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
1579 controller trees are mounted into it. Note that the
1580 directories mounted beneath it are not read-only. This is a
1581 security measure and is particularly useful because glibc
1582 actually includes a search logic to pick any tmpfs it can
1583 find to implement shm_open() if /dev/shm is not available
1584 (which it might very well be in namespaced setups).
1586 * machinectl gained a new "poweroff" command to cleanly power
1587 down a local OS container.
1589 * The PrivateDevices= unit file setting will now also drop the
1590 CAP_MKNOD capability from the capability bound set, and
1591 imply DevicePolicy=closed.
1593 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
1594 comprehensively on all long-running systemd services where
1595 this is appropriate.
1597 * systemd-udevd will now run in a disassociated mount
1598 namespace. To mount directories from udev rules, make sure to
1599 pull in mount units via SYSTEMD_WANTS properties.
1601 * The kdbus support gained support for uploading policy into
1602 the kernel. sd-bus gained support for creating "monitoring"
1603 connections that can eavesdrop into all bus communication
1604 for debugging purposes.
1606 * Timestamps may now be specified in seconds since the UNIX
1607 epoch Jan 1st, 1970 by specifying "@" followed by the value
1610 * Native tcpwrap support in systemd has been removed. tcpwrap
1611 is old code, not really maintained anymore and has serious
1612 shortcomings, and better options such as firewalls
1613 exist. For setups that require tcpwrap usage, please
1614 consider invoking your socket-activated service via tcpd,
1615 like on traditional inetd.
1617 * A new system.conf configuration option
1618 DefaultTimerAccuracySec= has been added that controls the
1619 default AccuracySec= setting of .timer units.
1621 * Timer units gained a new WakeSystem= switch. If enabled,
1622 timers configured this way will cause the system to resume
1623 from system suspend (if the system supports that, which most
1626 * Timer units gained a new Persistent= switch. If enabled,
1627 timers configured this way will save to disk when they have
1628 been last triggered. This information is then used on next
1629 reboot to possible execute overdue timer events, that
1630 could not take place because the system was powered off.
1631 This enables simple anacron-like behaviour for timer units.
1633 * systemctl's "list-timers" will now also list the time a
1634 timer unit was last triggered in addition to the next time
1635 it will be triggered.
1637 * systemd-networkd will now assign predictable IPv4LL
1638 addresses to its local interfaces.
1640 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
1641 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
1642 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
1643 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
1644 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
1645 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
1646 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
1647 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
1650 -- Berlin, 2014-03-25
1654 * A new unit file setting RestrictAddressFamilies= has been
1655 added to restrict which socket address families unit
1656 processes gain access to. This takes address family names
1657 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1658 attack surface of services via exotic protocol stacks. This
1659 is built on seccomp system call filters.
1661 * Two new unit file settings RuntimeDirectory= and
1662 RuntimeDirectoryMode= have been added that may be used to
1663 manage a per-daemon runtime directories below /run. This is
1664 an alternative for setting up directory permissions with
1665 tmpfiles snippets, and has the advantage that the runtime
1666 directory's lifetime is bound to the daemon runtime and that
1667 the daemon starts up with an empty directory each time. This
1668 is particularly useful when writing services that drop
1669 privileges using the User= or Group= setting.
1671 * The DeviceAllow= unit setting now supports globbing for
1672 matching against device group names.
1674 * The systemd configuration file system.conf gained new
1675 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1676 DefaultMemoryAccounting= to globally turn on/off accounting
1677 for specific resources (cgroups) for all units. These
1678 settings may still be overridden individually in each unit
1681 * systemd-gpt-auto-generator is now able to discover /srv and
1682 root partitions in addition to /home and swap partitions. It
1683 also supports LUKS-encrypted partitions now. With this in
1684 place, automatic discovery of partitions to mount following
1685 the Discoverable Partitions Specification
1686 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1687 is now a lot more complete. This allows booting without
1688 /etc/fstab and without root= on the kernel command line on
1689 systems prepared appropriately.
1691 * systemd-nspawn gained a new --image= switch which allows
1692 booting up disk images and Linux installations on any block
1693 device that follow the Discoverable Partitions Specification
1694 (see above). This means that installations made with
1695 appropriately updated installers may now be started and
1696 deployed using container managers, completely
1697 unmodified. (We hope that libvirt-lxc will add support for
1698 this feature soon, too.)
1700 * systemd-nspawn gained a new --network-macvlan= setting to
1701 set up a private macvlan interface for the
1702 container. Similarly, systemd-networkd gained a new
1703 Kind=macvlan setting in .netdev files.
1705 * systemd-networkd now supports configuring local addresses
1708 * A new tool systemd-network-wait-online has been added to
1709 synchronously wait for network connectivity using
1712 * The sd-bus.h bus API gained a new sd_bus_track object for
1713 tracking the life-cycle of bus peers. Note that sd-bus.h is
1714 still not a public API though (unless you specify
1715 --enable-kdbus on the configure command line, which however
1716 voids your warranty and you get no API stability guarantee).
1718 * The $XDG_RUNTIME_DIR runtime directories for each user are
1719 now individual tmpfs instances, which has the benefit of
1720 introducing separate pools for each user, with individual
1721 size limits, and thus making sure that unprivileged clients
1722 can no longer negatively impact the system or other users by
1723 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
1724 RuntimeDirectorySize= has been introduced that allows
1725 controlling the default size limit for all users. It
1726 defaults to 10% of the available physical memory. This is no
1727 replacement for quotas on tmpfs though (which the kernel
1728 still does not support), as /dev/shm and /tmp are still
1729 shared resources used by both the system and unprivileged
1732 * logind will now automatically turn off automatic suspending
1733 on laptop lid close when more than one display is
1734 connected. This was previously expected to be implemented
1735 individually in desktop environments (such as GNOME),
1736 however has been added to logind now, in order to fix a
1737 boot-time race where a desktop environment might not have
1738 been started yet and thus not been able to take an inhibitor
1739 lock at the time where logind already suspends the system
1740 due to a closed lid.
1742 * logind will now wait at least 30s after each system
1743 suspend/resume cycle, and 3min after system boot before
1744 suspending the system due to a closed laptop lid. This
1745 should give USB docking stations and similar enough time to
1746 be probed and configured after system resume and boot in
1747 order to then act as suspend blocker.
1749 * systemd-run gained a new --property= setting which allows
1750 initialization of resource control properties (and others)
1751 for the created scope or service unit. Example: "systemd-run
1752 --property=BlockIOWeight=10 updatedb" may be used to run
1753 updatedb at a low block IO scheduling weight.
1755 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1756 now also work in --scope mode.
1758 * When systemd is compiled with kdbus support, basic support
1759 for enforced policies is now in place. (Note that enabling
1760 kdbus still voids your warranty and no API compatibility
1763 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1764 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1765 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1766 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1767 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1768 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1769 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1770 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
1771 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
1772 Zbigniew Jędrzejewski-Szmek
1774 -- Berlin, 2014-03-12
1778 * systemd will now relabel /dev after loading the SMACK policy
1779 according to SMACK rules.
1781 * A new unit file option AppArmorProfile= has been added to
1782 set the AppArmor profile for the processes of a unit.
1784 * A new condition check ConditionArchitecture= has been added
1785 to conditionalize units based on the system architecture, as
1786 reported by uname()'s "machine" field.
1788 * systemd-networkd now supports matching on the system
1789 virtualization, architecture, kernel command line, host name
1792 * logind is now a lot more aggressive when suspending the
1793 machine due to a closed laptop lid. Instead of acting only
1794 on the lid close action, it will continuously watch the lid
1795 status and act on it. This is useful for laptops where the
1796 power button is on the outside of the chassis so that it can
1797 be reached without opening the lid (such as the Lenovo
1798 Yoga). On those machines, logind will now immediately
1799 re-suspend the machine if the power button has been
1800 accidentally pressed while the laptop was suspended and in a
1801 backpack or similar.
1803 * logind will now watch SW_DOCK switches and inhibit reaction
1804 to the lid switch if it is pressed. This means that logind
1805 will not suspend the machine anymore if the lid is closed
1806 and the system is docked, if the laptop supports SW_DOCK
1807 notifications via the input layer. Note that ACPI docking
1808 stations do not generate this currently. Also note that this
1809 logic is usually not fully sufficient and Desktop
1810 Environments should take a lid switch inhibitor lock when an
1811 external display is connected, as systemd will not watch
1814 * nspawn will now make use of the devices cgroup controller by
1815 default, and only permit creation of and access to the usual
1816 API device nodes like /dev/null or /dev/random, as well as
1817 access to (but not creation of) the pty devices.
1819 * We will now ship a default .network file for
1820 systemd-networkd that automatically configures DHCP for
1821 network interfaces created by nspawn's --network-veth or
1822 --network-bridge= switches.
1824 * systemd will now understand the usual M, K, G, T suffixes
1825 according to SI conventions (i.e. to the base 1000) when
1826 referring to throughput and hardware metrics. It will stay
1827 with IEC conventions (i.e. to the base 1024) for software
1828 metrics, according to what is customary according to
1829 Wikipedia. We explicitly document which base applies for
1830 each configuration option.
1832 * The DeviceAllow= setting in unit files now supports a syntax
1833 to whitelist an entire group of devices node majors at once,
1834 based on the /proc/devices listing. For example, with the
1835 string "char-pts", it is now possible to whitelist all
1836 current and future pseudo-TTYs at once.
1838 * sd-event learned a new "post" event source. Event sources of
1839 this type are triggered by the dispatching of any event
1840 source of a type that is not "post". This is useful for
1841 implementing clean-up and check event sources that are
1842 triggered by other work being done in the program.
1844 * systemd-networkd is no longer statically enabled, but uses
1845 the usual [Install] sections so that it can be
1846 enabled/disabled using systemctl. It still is enabled by
1849 * When creating a veth interface pair with systemd-nspawn, the
1850 host side will now be prefixed with "vb-" if
1851 --network-bridge= is used, and with "ve-" if --network-veth
1852 is used. This way, it is easy to distinguish these cases on
1853 the host, for example to apply different configuration to
1854 them with systemd-networkd.
1856 * The compatibility libraries for libsystemd-journal.so,
1857 libsystem-id128.so, libsystemd-login.so and
1858 libsystemd-daemon.so do not make use of IFUNC
1859 anymore. Instead, we now build libsystemd.so multiple times
1860 under these alternative names. This means that the footprint
1861 is drastically increased, but given that these are
1862 transitional compatibility libraries, this should not matter
1863 much. This change has been made necessary to support the ARM
1864 platform for these compatibility libraries, as the ARM
1865 toolchain is not really at the same level as the toolchain
1866 for other architectures like x86 and does not support
1867 IFUNC. Please make sure to use --enable-compat-libs only
1868 during a transitional period!
1870 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
1871 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1872 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
1873 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
1874 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
1875 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
1876 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
1877 Zbigniew Jędrzejewski-Szmek
1879 -- Berlin, 2014-02-24
1883 * A new component "systemd-networkd" has been added that can
1884 be used to configure local network interfaces statically or
1885 via DHCP. It is capable of bringing up bridges, VLANs, and
1886 bonding. Currently, no hook-ups for interactive network
1887 configuration are provided. Use this for your initrd,
1888 container, embedded, or server setup if you need a simple,
1889 yet powerful, network configuration solution. This
1890 configuration subsystem is quite nifty, as it allows wildcard
1891 hotplug matching in interfaces. For example, with a single
1892 configuration snippet, you can configure that all Ethernet
1893 interfaces showing up are automatically added to a bridge,
1894 or similar. It supports link-sensing and more.
1896 * A new tool "systemd-socket-proxyd" has been added which can
1897 act as a bidirectional proxy for TCP sockets. This is
1898 useful for adding socket activation support to services that
1899 do not actually support socket activation, including virtual
1900 machines and the like.
1902 * Add a new tool to save/restore rfkill state on
1905 * Save/restore state of keyboard backlights in addition to
1906 display backlights on shutdown/boot.
1908 * udev learned a new SECLABEL{} construct to label device
1909 nodes with a specific security label when they appear. For
1910 now, only SECLABEL{selinux} is supported, but the syntax is
1911 prepared for additional security frameworks.
1913 * udev gained a new scheme to configure link-level attributes
1914 from files in /etc/systemd/network/*.link. These files can
1915 match against MAC address, device path, driver name and type,
1916 and will apply attributes like the naming policy, link speed,
1917 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
1918 address assignment policy (randomized, ...).
1920 * The configuration of network interface naming rules for
1921 "permanent interface names" has changed: a new NamePolicy=
1922 setting in the [Link] section of .link files determines the
1923 priority of possible naming schemes (onboard, slot, mac,
1924 path). The default value of this setting is determined by
1925 /usr/lib/net/links/99-default.link. Old
1926 80-net-name-slot.rules udev configuration file has been
1927 removed, so local configuration overriding this file should
1928 be adapated to override 99-default.link instead.
1930 * When the User= switch is used in a unit file, also
1931 initialize $SHELL= based on the user database entry.
1933 * systemd no longer depends on libdbus. All communication is
1934 now done with sd-bus, systemd's low-level bus library
1937 * kdbus support has been added to PID 1 itself. When kdbus is
1938 enabled, this causes PID 1 to set up the system bus and
1939 enable support for a new ".busname" unit type that
1940 encapsulates bus name activation on kdbus. It works a little
1941 bit like ".socket" units, except for bus names. A new
1942 generator has been added that converts classic dbus1 service
1943 activation files automatically into native systemd .busname
1946 * sd-bus: add a light-weight vtable implementation that allows
1947 defining objects on the bus with a simple static const
1948 vtable array of its methods, signals and properties.
1950 * systemd will not generate or install static dbus
1951 introspection data anymore to /usr/share/dbus-1/interfaces,
1952 as the precise format of these files is unclear, and
1953 nothing makes use of it.
1955 * A proxy daemon is now provided to proxy clients connecting
1956 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1957 compatibility with classic D-Bus.
1959 * A bus driver implementation has been added that supports the
1960 classic D-Bus bus driver calls on kdbus, also for
1961 compatibility purposes.
1963 * A new API "sd-event.h" has been added that implements a
1964 minimal event loop API built around epoll. It provides a
1965 couple of features that direct epoll usage is lacking:
1966 prioritization of events, scales to large numbers of timer
1967 events, per-event timer slack (accuracy), system-wide
1968 coalescing of timer events, exit handlers, watchdog
1969 supervision support using systemd's sd_notify() API, child
1972 * A new API "sd-rntl.h" has been added that provides an API
1973 around the route netlink interface of the kernel, similar in
1974 style to "sd-bus.h".
1976 * A new API "sd-dhcp-client.h" has been added that provides a
1977 small DHCPv4 client-side implementation. This is used by
1980 * There is a new kernel command line option
1981 "systemd.restore_state=0|1". When set to "0", none of the
1982 systemd tools will restore saved runtime state to hardware
1983 devices. More specifically, the rfkill and backlight states
1986 * The FsckPassNo= compatibility option in mount/service units
1987 has been removed. The fstab generator will now add the
1988 necessary dependencies automatically, and does not require
1989 PID1's support for that anymore.
1991 * journalctl gained a new switch, --list-boots, that lists
1992 recent boots with their times and boot IDs.
1994 * The various tools like systemctl, loginctl, timedatectl,
1995 busctl, systemd-run, ... have gained a new switch "-M" to
1996 connect to a specific, local OS container (as direct
1997 connection, without requiring SSH). This works on any
1998 container that is registered with machined, such as those
1999 created by libvirt-lxc or nspawn.
2001 * systemd-run and systemd-analyze also gained support for "-H"
2002 to connect to remote hosts via SSH. This is particularly
2003 useful for systemd-run because it enables queuing of jobs
2004 onto remote systems.
2006 * machinectl gained a new command "login" to open a getty
2007 login in any local container. This works with any container
2008 that is registered with machined (such as those created by
2009 libvirt-lxc or nspawn), and which runs systemd inside.
2011 * machinectl gained a new "reboot" command that may be used to
2012 trigger a reboot on a specific container that is registered
2013 with machined. This works on any container that runs an init
2014 system of some kind.
2016 * systemctl gained a new "list-timers" command to print a nice
2017 listing of installed timer units with the times they elapse
2020 * Alternative reboot() parameters may now be specified on the
2021 "systemctl reboot" command line and are passed to the
2022 reboot() system call.
2024 * systemctl gained a new --job-mode= switch to configure the
2025 mode to queue a job with. This is a more generic version of
2026 --fail, --irreversible, and --ignore-dependencies, which are
2027 still available but not advertised anymore.
2029 * /etc/systemd/system.conf gained new settings to configure
2030 various default timeouts of units, as well as the default
2031 start limit interval and burst. These may still be overridden
2034 * PID1 will now export on the bus profile data of the security
2035 policy upload process (such as the SELinux policy upload to
2038 * journald: when forwarding logs to the console, include
2039 timestamps (following the setting in
2040 /sys/module/printk/parameters/time).
2042 * OnCalendar= in timer units now understands the special
2043 strings "yearly" and "annually". (Both are equivalent)
2045 * The accuracy of timer units is now configurable with the new
2046 AccuracySec= setting. It defaults to 1min.
2048 * A new dependency type JoinsNamespaceOf= has been added that
2049 allows running two services within the same /tmp and network
2050 namespace, if PrivateNetwork= or PrivateTmp= are used.
2052 * A new command "cat" has been added to systemctl. It outputs
2053 the original unit file of a unit, and concatenates the
2054 contents of additional "drop-in" unit file snippets, so that
2055 the full configuration is shown.
2057 * systemctl now supports globbing on the various "list-xyz"
2058 commands, like "list-units" or "list-sockets", as well as on
2059 those commands which take multiple unit names.
2061 * journalctl's --unit= switch gained support for globbing.
2063 * All systemd daemons now make use of the watchdog logic so
2064 that systemd automatically notices when they hang.
2066 * If the $container_ttys environment variable is set,
2067 getty-generator will automatically spawn a getty for each
2068 listed tty. This is useful for container managers to request
2069 login gettys to be spawned on as many ttys as needed.
2071 * %h, %s, %U specifier support is not available anymore when
2072 used in unit files for PID 1. This is because NSS calls are
2073 not safe from PID 1. They stay available for --user
2074 instances of systemd, and as special case for the root user.
2076 * loginctl gained a new "--no-legend" switch to turn off output
2079 * The "sd-login.h" API gained three new calls:
2080 sd_session_is_remote(), sd_session_get_remote_user(),
2081 sd_session_get_remote_host() to query information about
2084 * The udev hardware database now also carries vendor/product
2085 information of SDIO devices.
2087 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
2088 determine whether watchdog notifications are requested by
2091 * Socket-activated per-connection services now include a
2092 short description of the connection parameters in the
2095 * tmpfiles gained a new "--boot" option. When this is not used,
2096 only lines where the command character is not suffixed with
2097 "!" are executed. When this option is specified, those
2098 options are executed too. This partitions tmpfiles
2099 directives into those that can be safely executed at any
2100 time, and those which should be run only at boot (for
2101 example, a line that creates /run/nologin).
2103 * A new API "sd-resolve.h" has been added which provides a simple
2104 asynchronous wrapper around glibc NSS host name resolution
2105 calls, such as getaddrinfo(). In contrast to glibc's
2106 getaddrinfo_a(), it does not use signals. In contrast to most
2107 other asynchronous name resolution libraries, this one does
2108 not reimplement DNS, but reuses NSS, so that alternate
2109 host name resolution systems continue to work, such as mDNS,
2110 LDAP, etc. This API is based on libasyncns, but it has been
2111 cleaned up for inclusion in systemd.
2113 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
2114 "sd-daemon.h" are no longer found in individual libraries
2115 libsystemd-journal.so, libsystemd-login.so,
2116 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
2117 merged them into a single library, libsystemd.so, which
2118 provides all symbols. The reason for this is cyclic
2119 dependencies, as these libraries tend to use each other's
2120 symbols. So far, we have managed to workaround that by linking
2121 a copy of a good part of our code into each of these
2122 libraries again and again, which, however, makes certain
2123 things hard to do, like sharing static variables. Also, it
2124 substantially increases footprint. With this change, there
2125 is only one library for the basic APIs systemd
2126 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
2127 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
2128 library as well, however are subject to the --enable-kdbus
2129 switch (see below). Note that "sd-dhcp-client.h" is not part
2130 of this library (this is because it only consumes, never
2131 provides, services of/to other APIs). To make the transition
2132 easy from the separate libraries to the unified one, we
2133 provide the --enable-compat-libs compile-time switch which
2134 will generate stub libraries that are compatible with the
2135 old ones but redirect all calls to the new one.
2137 * All of the kdbus logic and the new APIs "sd-bus.h",
2138 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
2139 and "sd-utf8.h" are compile-time optional via the
2140 "--enable-kdbus" switch, and they are not compiled in by
2141 default. To make use of kdbus, you have to explicitly enable
2142 the switch. Note however, that neither the kernel nor the
2143 userspace API for all of this is considered stable yet. We
2144 want to maintain the freedom to still change the APIs for
2145 now. By specifying this build-time switch, you acknowledge
2146 that you are aware of the instability of the current
2149 * Also, note that while kdbus is pretty much complete,
2150 it lacks one thing: proper policy support. This means you
2151 can build a fully working system with all features; however,
2152 it will be highly insecure. Policy support will be added in
2153 one of the next releases, at the same time that we will
2154 declare the APIs stable.
2156 * When the kernel command line argument "kdbus" is specified,
2157 systemd will automatically load the kdbus.ko kernel module. At
2158 this stage of development, it is only useful for testing kdbus
2159 and should not be used in production. Note: if "--enable-kdbus"
2160 is specified, and the kdbus.ko kernel module is available, and
2161 "kdbus" is added to the kernel command line, the entire system
2162 runs with kdbus instead of dbus-daemon, with the above mentioned
2163 problem of missing the system policy enforcement. Also a future
2164 version of kdbus.ko or a newer systemd will not be compatible with
2165 each other, and will unlikely be able to boot the machine if only
2166 one of them is updated.
2168 * systemctl gained a new "import-environment" command which
2169 uploads the caller's environment (or parts thereof) into the
2170 service manager so that it is inherited by services started
2171 by the manager. This is useful to upload variables like
2172 $DISPLAY into the user service manager.
2174 * A new PrivateDevices= switch has been added to service units
2175 which allows running a service with a namespaced /dev
2176 directory that does not contain any device nodes for
2177 physical devices. More specifically, it only includes devices
2178 such as /dev/null, /dev/urandom, and /dev/zero which are API
2181 * logind has been extended to support behaviour like VT
2182 switching on seats that do not support a VT. This makes
2183 multi-session available on seats that are not the first seat
2184 (seat0), and on systems where kernel support for VTs has
2185 been disabled at compile-time.
2187 * If a process holds a delay lock for system sleep or shutdown
2188 and fails to release it in time, we will now log its
2189 identity. This makes it easier to identify processes that
2190 cause slow suspends or power-offs.
2192 * When parsing /etc/crypttab, support for a new key-slot=
2193 option as supported by Debian is added. It allows indicating
2194 which LUKS slot to use on disk, speeding up key loading.
2196 * The sd_journald_sendv() API call has been checked and
2197 officially declared to be async-signal-safe so that it may
2198 be invoked from signal handlers for logging purposes.
2200 * Boot-time status output is now enabled automatically after a
2201 short timeout if boot does not progress, in order to give
2202 the user an indication what she or he is waiting for.
2204 * The boot-time output has been improved to show how much time
2205 remains until jobs expire.
2207 * The KillMode= switch in service units gained a new possible
2208 value "mixed". If set, and the unit is shut down, then the
2209 initial SIGTERM signal is sent only to the main daemon
2210 process, while the following SIGKILL signal is sent to
2211 all remaining processes of the service.
2213 * When a scope unit is registered, a new property "Controller"
2214 may be set. If set to a valid bus name, systemd will send a
2215 RequestStop() signal to this name when it would like to shut
2216 down the scope. This may be used to hook manager logic into
2217 the shutdown logic of scope units. Also, scope units may now
2218 be put in a special "abandoned" state, in which case the
2219 manager process which created them takes no further
2220 responsibilities for it.
2222 * When reading unit files, systemd will now verify
2223 the access mode of these files, and warn about certain
2224 suspicious combinations. This has been added to make it
2225 easier to track down packaging bugs where unit files are
2226 marked executable or world-writable.
2228 * systemd-nspawn gained a new "--setenv=" switch to set
2229 container-wide environment variables. The similar option in
2230 systemd-activate was renamed from "--environment=" to
2231 "--setenv=" for consistency.
2233 * systemd-nspawn has been updated to create a new kdbus domain
2234 for each container that is invoked, thus allowing each
2235 container to have its own set of system and user buses,
2236 independent of the host.
2238 * systemd-nspawn gained a new --drop-capability= switch to run
2239 the container with less capabilities than the default. Both
2240 --drop-capability= and --capability= now take the special
2241 string "all" for dropping or keeping all capabilities.
2243 * systemd-nspawn gained new switches for executing containers
2244 with specific SELinux labels set.
2246 * systemd-nspawn gained a new --quiet switch to not generate
2247 any additional output but the container's own console
2250 * systemd-nspawn gained a new --share-system switch to run a
2251 container without PID namespacing enabled.
2253 * systemd-nspawn gained a new --register= switch to control
2254 whether the container is registered with systemd-machined or
2255 not. This is useful for containers that do not run full
2256 OS images, but only specific apps.
2258 * systemd-nspawn gained a new --keep-unit which may be used
2259 when invoked as the only program from a service unit, and
2260 results in registration of the unit service itself in
2261 systemd-machined, instead of a newly opened scope unit.
2263 * systemd-nspawn gained a new --network-interface= switch for
2264 moving arbitrary interfaces to the container. The new
2265 --network-veth switch creates a virtual Ethernet connection
2266 between host and container. The new --network-bridge=
2267 switch then allows assigning the host side of this virtual
2268 Ethernet connection to a bridge device.
2270 * systemd-nspawn gained a new --personality= switch for
2271 setting the kernel personality for the container. This is
2272 useful when running a 32-bit container on a 64-bit host. A
2273 similar option Personality= is now also available for service
2276 * logind will now also track a "Desktop" identifier for each
2277 session which encodes the desktop environment of it. This is
2278 useful for desktop environments that want to identify
2279 multiple running sessions of itself easily.
2281 * A new SELinuxContext= setting for service units has been
2282 added that allows setting a specific SELinux execution
2283 context for a service.
2285 * Most systemd client tools will now honour $SYSTEMD_LESS for
2286 settings of the "less" pager. By default, these tools will
2287 override $LESS to allow certain operations to work, such as
2288 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
2289 influence this logic.
2291 * systemd's "seccomp" hook-up has been changed to make use of
2292 the libseccomp library instead of using its own
2293 implementation. This has benefits for portability among
2296 * For usage together with SystemCallFilter=, a new
2297 SystemCallErrorNumber= setting has been introduced that
2298 allows configuration of a system error number to be returned
2299 on filtered system calls, instead of immediately killing the
2300 process. Also, SystemCallArchitectures= has been added to
2301 limit access to system calls of a particular architecture
2302 (in order to turn off support for unused secondary
2303 architectures). There is also a global
2304 SystemCallArchitectures= setting in system.conf now to turn
2305 off support for non-native system calls system-wide.
2307 * systemd requires a kernel with a working name_to_handle_at(),
2308 please see the kernel config requirements in the README file.
2310 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
2311 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
2312 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
2313 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
2314 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
2315 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
2316 Elia Pinto, Florian Weimer, George McCollister, Goffredo
2317 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
2318 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
2319 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
2320 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
2321 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
2322 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
2323 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
2324 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
2325 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
2326 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
2327 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
2328 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
2329 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
2330 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
2331 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
2332 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
2333 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
2335 -- Berlin, 2014-02-20
2339 * logind has gained support for facilitating privileged input
2340 and drm device access for unprivileged clients. This work is
2341 useful to allow Wayland display servers (and similar
2342 programs, such as kmscon) to run under the user's ID and
2343 access input and drm devices which are normally
2344 protected. When this is used (and the kernel is new enough)
2345 logind will "mute" IO on the file descriptors passed to
2346 Wayland as long as it is in the background and "unmute" it
2347 if it returns into the foreground. This allows secure
2348 session switching without allowing background sessions to
2349 eavesdrop on input and display data. This also introduces
2350 session switching support if VT support is turned off in the
2351 kernel, and on seats that are not seat0.
2353 * A new kernel command line option luks.options= is understood
2354 now which allows specifying LUKS options for usage for LUKS
2355 encrypted partitions specified with luks.uuid=.
2357 * tmpfiles.d(5) snippets may now use specifier expansion in
2358 path names. More specifically %m, %b, %H, %v, are now
2359 replaced by the local machine id, boot id, hostname, and
2360 kernel version number.
2362 * A new tmpfiles.d(5) command "m" has been introduced which
2363 may be used to change the owner/group/access mode of a file
2364 or directory if it exists, but do nothing if it does not.
2366 * This release removes high-level support for the
2367 MemorySoftLimit= cgroup setting. The underlying kernel
2368 cgroup attribute memory.soft_limit= is currently badly
2369 designed and likely to be removed from the kernel API in its
2370 current form, hence we should not expose it for now.
2372 * The memory.use_hierarchy cgroup attribute is now enabled for
2373 all cgroups systemd creates in the memory cgroup
2374 hierarchy. This option is likely to be come the built-in
2375 default in the kernel anyway, and the non-hierarchical mode
2376 never made much sense in the intrinsically hierarchical
2379 * A new field _SYSTEMD_SLICE= is logged along with all journal
2380 messages containing the slice a message was generated
2381 from. This is useful to allow easy per-customer filtering of
2382 logs among other things.
2384 * systemd-journald will no longer adjust the group of journal
2385 files it creates to the "systemd-journal" group. Instead we
2386 rely on the journal directory to be owned by the
2387 "systemd-journal" group, and its setgid bit set, so that the
2388 kernel file system layer will automatically enforce that
2389 journal files inherit this group assignment. The reason for
2390 this change is that we cannot allow NSS look-ups from
2391 journald which would be necessary to resolve
2392 "systemd-journal" to a numeric GID, because this might
2393 create deadlocks if NSS involves synchronous queries to
2394 other daemons (such as nscd, or sssd) which in turn are
2395 logging clients of journald and might block on it, which
2396 would then dead lock. A tmpfiles.d(5) snippet included in
2397 systemd will make sure the setgid bit and group are
2398 properly set on the journal directory if it exists on every
2399 boot. However, we recommend adjusting it manually after
2400 upgrades too (or from RPM scriptlets), so that the change is
2401 not delayed until next reboot.
2403 * Backlight and random seed files in /var/lib/ have moved into
2404 the /var/lib/systemd/ directory, in order to centralize all
2405 systemd generated files in one directory.
2407 * Boot time performance measurements (as displayed by
2408 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
2409 performance information if that's available to determine how
2410 much time BIOS and boot loader initialization required. With
2411 a sufficiently new BIOS you hence no longer need to boot
2412 with Gummiboot to get access to such information.
2414 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
2415 Cristian Rodríguez, Dave Reisner, David Herrmann, David
2416 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
2417 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
2418 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
2419 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
2420 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2422 -- Berlin, 2013-10-02
2426 * The Restart= option for services now understands a new
2427 on-watchdog setting, which will restart the service
2428 automatically if the service stops sending out watchdog keep
2429 alive messages (as configured with WatchdogSec=).
2431 * The getty generator (which is responsible for bringing up a
2432 getty on configured serial consoles) will no longer only
2433 start a getty on the primary kernel console but on all
2434 others, too. This makes the order in which console= is
2435 specified on the kernel command line less important.
2437 * libsystemd-logind gained a new sd_session_get_vt() call to
2438 retrieve the VT number of a session.
2440 * If the option "tries=0" is set for an entry of /etc/crypttab
2441 its passphrase is queried indefinitely instead of any
2442 maximum number of tries.
2444 * If a service with a configure PID file terminates its PID
2445 file will now be removed automatically if it still exists
2446 afterwards. This should put an end to stale PID files.
2448 * systemd-run will now also take relative binary path names
2449 for execution and no longer insists on absolute paths.
2451 * InaccessibleDirectories= and ReadOnlyDirectories= now take
2452 paths that are optionally prefixed with "-" to indicate that
2453 it should not be considered a failure if they do not exist.
2455 * journalctl -o (and similar commands) now understands a new
2456 output mode "short-precise", it is similar to "short" but
2457 shows timestamps with usec accuracy.
2459 * The option "discard" (as known from Debian) is now
2460 synonymous to "allow-discards" in /etc/crypttab. In fact,
2461 "discard" is preferred now (since it is easier to remember
2464 * Some licensing clean-ups were made, so that more code is now
2465 LGPL-2.1 licensed than before.
2467 * A minimal tool to save/restore the display backlight
2468 brightness across reboots has been added. It will store the
2469 backlight setting as late as possible at shutdown, and
2470 restore it as early as possible during reboot.
2472 * A logic to automatically discover and enable home and swap
2473 partitions on GPT disks has been added. With this in place
2474 /etc/fstab becomes optional for many setups as systemd can
2475 discover certain partitions located on the root disk
2476 automatically. Home partitions are recognized under their
2477 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
2478 partitions are recognized under their GPT type ID
2479 0657fd6da4ab43c484e50933c84b4f4f.
2481 * systemd will no longer pass any environment from the kernel
2482 or initrd to system services. If you want to set an
2483 environment for all services, do so via the kernel command
2484 line systemd.setenv= assignment.
2486 * The systemd-sysctl tool no longer natively reads the file
2487 /etc/sysctl.conf. If desired, the file should be symlinked
2488 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
2489 legacy support by a symlink rather than built-in code, it
2490 also makes the otherwise hidden order of application of the
2491 different files visible. (Note that this partly reverts to a
2492 pre-198 application order of sysctl knobs!)
2494 * The "systemctl set-log-level" and "systemctl dump" commands
2495 have been moved to systemd-analyze.
2497 * systemd-run learned the new --remain-after-exit switch,
2498 which causes the scope unit not to be cleaned up
2499 automatically after the process terminated.
2501 * tmpfiles learned a new --exclude-prefix= switch to exclude
2502 certain paths from operation.
2504 * journald will now automatically flush all messages to disk
2505 as soon as a message at the log level CRIT, ALERT or EMERG
2508 Contributions from: Andrew Cook, Brandon Philips, Christian
2509 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
2510 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
2511 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
2512 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
2513 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
2514 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
2515 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
2516 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
2517 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
2518 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
2519 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
2520 William Giokas, Zbigniew Jędrzejewski-Szmek
2522 -- Berlin, 2013-09-13
2526 * The documentation has been updated to cover the various new
2527 concepts introduced with 205.
2529 * Unit files now understand the new %v specifier which
2530 resolves to the kernel version string as returned by "uname
2533 * systemctl now supports filtering the unit list output by
2534 load state, active state and sub state, using the new
2537 * "systemctl status" will now show the results of the
2538 condition checks (like ConditionPathExists= and similar) of
2539 the last start attempts of the unit. They are also logged to
2542 * "journalctl -b" may now be used to look for boot output of a
2543 specific boot. Try "journalctl -b -1" for the previous boot,
2544 but the syntax is substantially more powerful.
2546 * "journalctl --show-cursor" has been added which prints the
2547 cursor string the last shown log line. This may then be used
2548 with the new "journalctl --after-cursor=" switch to continue
2549 browsing logs from that point on.
2551 * "journalctl --force" may now be used to force regeneration
2554 * Creation of "dead" device nodes has been moved from udev
2555 into kmod and tmpfiles. Previously, udev would read the kmod
2556 databases to pre-generate dead device nodes based on meta
2557 information contained in kernel modules, so that these would
2558 be auto-loaded on access rather then at boot. As this
2559 does not really have much to do with the exposing actual
2560 kernel devices to userspace this has always been slightly
2561 alien in the udev codebase. Following the new scheme kmod
2562 will now generate a runtime snippet for tmpfiles from the
2563 module meta information and it now is tmpfiles' job to the
2564 create the nodes. This also allows overriding access and
2565 other parameters for the nodes using the usual tmpfiles
2566 facilities. As side effect this allows us to remove the
2567 CAP_SYS_MKNOD capability bit from udevd entirely.
2569 * logind's device ACLs may now be applied to these "dead"
2570 devices nodes too, thus finally allowing managed access to
2571 devices such as /dev/snd/sequencer whithout loading the
2572 backing module right-away.
2574 * A new RPM macro has been added that may be used to apply
2575 tmpfiles configuration during package installation.
2577 * systemd-detect-virt and ConditionVirtualization= now can
2578 detect User-Mode-Linux machines (UML).
2580 * journald will now implicitly log the effective capabilities
2581 set of processes in the message metadata.
2583 * systemd-cryptsetup has gained support for TrueCrypt volumes.
2585 * The initrd interface has been simplified (more specifically,
2586 support for passing performance data via environment
2587 variables and fsck results via files in /run has been
2588 removed). These features were non-essential, and are
2589 nowadays available in a much nicer way by having systemd in
2590 the initrd serialize its state and have the hosts systemd
2591 deserialize it again.
2593 * The udev "keymap" data files and tools to apply keyboard
2594 specific mappings of scan to key codes, and force-release
2595 scan code lists have been entirely replaced by a udev
2596 "keyboard" builtin and a hwdb data file.
2598 * systemd will now honour the kernel's "quiet" command line
2599 argument also during late shutdown, resulting in a
2600 completely silent shutdown when used.
2602 * There's now an option to control the SO_REUSEPORT socket
2603 option in .socket units.
2605 * Instance units will now automatically get a per-template
2606 subslice of system.slice unless something else is explicitly
2607 configured. For example, instances of sshd@.service will now
2608 implicitly be placed in system-sshd.slice rather than
2609 system.slice as before.
2611 * Test coverage support may now be enabled at build time.
2613 Contributions from: Dave Reisner, Frederic Crozat, Harald
2614 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
2615 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
2616 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
2617 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
2618 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
2619 Giokas, Zbigniew Jędrzejewski-Szmek
2621 -- Berlin, 2013-07-23
2625 * Two new unit types have been introduced:
2627 Scope units are very similar to service units, however, are
2628 created out of pre-existing processes -- instead of PID 1
2629 forking off the processes. By using scope units it is
2630 possible for system services and applications to group their
2631 own child processes (worker processes) in a powerful way
2632 which then maybe used to organize them, or kill them
2633 together, or apply resource limits on them.
2635 Slice units may be used to partition system resources in an
2636 hierarchical fashion and then assign other units to them. By
2637 default there are now three slices: system.slice (for all
2638 system services), user.slice (for all user sessions),
2639 machine.slice (for VMs and containers).
2641 Slices and scopes have been introduced primarily in
2642 context of the work to move cgroup handling to a
2643 single-writer scheme, where only PID 1
2644 creates/removes/manages cgroups.
2646 * There's a new concept of "transient" units. In contrast to
2647 normal units these units are created via an API at runtime,
2648 not from configuration from disk. More specifically this
2649 means it is now possible to run arbitrary programs as
2650 independent services, with all execution parameters passed
2651 in via bus APIs rather than read from disk. Transient units
2652 make systemd substantially more dynamic then it ever was,
2653 and useful as a general batch manager.
2655 * logind has been updated to make use of scope and slice units
2656 for managing user sessions. As a user logs in he will get
2657 his own private slice unit, to which all sessions are added
2658 as scope units. We also added support for automatically
2659 adding an instance of user@.service for the user into the
2660 slice. Effectively logind will no longer create cgroup
2661 hierarchies on its own now, it will defer entirely to PID 1
2662 for this by means of scope, service and slice units. Since
2663 user sessions this way become entities managed by PID 1
2664 the output of "systemctl" is now a lot more comprehensive.
2666 * A new mini-daemon "systemd-machined" has been added which
2667 may be used by virtualization managers to register local
2668 VMs/containers. nspawn has been updated accordingly, and
2669 libvirt will be updated shortly. machined will collect a bit
2670 of meta information about the VMs/containers, and assign
2671 them their own scope unit (see above). The collected
2672 meta-data is then made available via the "machinectl" tool,
2673 and exposed in "ps" and similar tools. machined/machinectl
2674 is compile-time optional.
2676 * As discussed earlier, the low-level cgroup configuration
2677 options ControlGroup=, ControlGroupModify=,
2678 ControlGroupPersistent=, ControlGroupAttribute= have been
2679 removed. Please use high-level attribute settings instead as
2680 well as slice units.
2682 * A new bus call SetUnitProperties() has been added to alter
2683 various runtime parameters of a unit. This is primarily
2684 useful to alter cgroup parameters dynamically in a nice way,
2685 but will be extended later on to make more properties
2686 modifiable at runtime. systemctl gained a new set-properties
2687 command that wraps this call.
2689 * A new tool "systemd-run" has been added which can be used to
2690 run arbitrary command lines as transient services or scopes,
2691 while configuring a number of settings via the command
2692 line. This tool is currently very basic, however already
2693 very useful. We plan to extend this tool to even allow
2694 queuing of execution jobs with time triggers from the
2695 command line, similar in fashion to "at".
2697 * nspawn will now inform the user explicitly that kernels with
2698 audit enabled break containers, and suggest the user to turn
2701 * Support for detecting the IMA and AppArmor security
2702 frameworks with ConditionSecurity= has been added.
2704 * journalctl gained a new "-k" switch for showing only kernel
2705 messages, mimicking dmesg output; in addition to "--user"
2706 and "--system" switches for showing only user's own logs
2709 * systemd-delta can now show information about drop-in
2710 snippets extending unit files.
2712 * libsystemd-bus has been substantially updated but is still
2713 not available as public API.
2715 * systemd will now look for the "debug" argument on the kernel
2716 command line and enable debug logging, similar to what
2717 "systemd.log_level=debug" already did before.
2719 * "systemctl set-default", "systemctl get-default" has been
2720 added to configure the default.target symlink, which
2721 controls what to boot into by default.
2723 * "systemctl set-log-level" has been added as a convenient
2724 way to raise and lower systemd logging threshold.
2726 * "systemd-analyze plot" will now show the time the various
2727 generators needed for execution, as well as information
2728 about the unit file loading.
2730 * libsystemd-journal gained a new sd_journal_open_files() call
2731 for opening specific journal files. journactl also gained a
2732 new switch to expose this new functionality. Previously we
2733 only supported opening all files from a directory, or all
2734 files from the system, as opening individual files only is
2735 racy due to journal file rotation.
2737 * systemd gained the new DefaultEnvironment= setting in
2738 /etc/systemd/system.conf to set environment variables for
2741 * If a privileged process logs a journal message with the
2742 OBJECT_PID= field set, then journald will automatically
2743 augment this with additional OBJECT_UID=, OBJECT_GID=,
2744 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2745 system services want to log events about specific client
2746 processes. journactl/systemctl has been updated to make use
2747 of this information if all log messages regarding a specific
2750 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2751 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2752 Reisner, David Coppa, David King, David Strauss, Eelco
2753 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2754 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2755 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2756 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2757 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2758 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2759 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2760 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2761 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2762 Łukasz Stelmach, 장동준
2766 * The Python bindings gained some minimal support for the APIs
2767 exposed by libsystemd-logind.
2769 * ConditionSecurity= gained support for detecting SMACK. Since
2770 this condition already supports SELinux and AppArmor we only
2771 miss IMA for this. Patches welcome!
2773 Contributions from: Karol Lewandowski, Lennart Poettering,
2774 Zbigniew Jędrzejewski-Szmek
2778 * systemd-nspawn will now create /etc/resolv.conf if
2779 necessary, before bind-mounting the host's file onto it.
2781 * systemd-nspawn will now store meta information about a
2782 container on the container's cgroup as extended attribute
2783 fields, including the root directory.
2785 * The cgroup hierarchy has been reworked in many ways. All
2786 objects any of the components systemd creates in the cgroup
2787 tree are now suffixed. More specifically, user sessions are
2788 now placed in cgroups suffixed with ".session", users in
2789 cgroups suffixed with ".user", and nspawn containers in
2790 cgroups suffixed with ".nspawn". Furthermore, all cgroup
2791 names are now escaped in a simple scheme to avoid collision
2792 of userspace object names with kernel filenames. This work
2793 is preparation for making these objects relocatable in the
2794 cgroup tree, in order to allow easy resource partitioning of
2795 these objects without causing naming conflicts.
2797 * systemctl list-dependencies gained the new switches
2798 --plain, --reverse, --after and --before.
2800 * systemd-inhibit now shows the process name of processes that
2801 have taken an inhibitor lock.
2803 * nss-myhostname will now also resolve "localhost"
2804 implicitly. This makes /etc/hosts an optional file and
2805 nicely handles that on IPv6 ::1 maps to both "localhost" and
2808 * libsystemd-logind.so gained a new call
2809 sd_get_machine_names() to enumerate running containers and
2810 VMs (currently only supported by very new libvirt and
2811 nspawn). sd_login_monitor can now be used to watch
2812 VMs/containers coming and going.
2814 * .include is not allowed recursively anymore, and only in
2815 unit files. Usually it is better to use drop-in snippets in
2816 .d/*.conf anyway, as introduced with systemd 198.
2818 * systemd-analyze gained a new "critical-chain" command that
2819 determines the slowest chain of units run during system
2820 boot-up. It is very useful for tracking down where
2821 optimizing boot time is the most beneficial.
2823 * systemd will no longer allow manipulating service paths in
2824 the name=systemd:/system cgroup tree using ControlGroup= in
2825 units. (But is still fine with it in all other dirs.)
2827 * There's a new systemd-nspawn@.service service file that may
2828 be used to easily run nspawn containers as system
2829 services. With the container's root directory in
2830 /var/lib/container/foobar it is now sufficient to run
2831 "systemctl start systemd-nspawn@foobar.service" to boot it.
2833 * systemd-cgls gained a new parameter "--machine" to list only
2834 the processes within a certain container.
2836 * ConditionSecurity= now can check for "apparmor". We still
2837 are lacking checks for SMACK and IMA for this condition
2838 check though. Patches welcome!
2840 * A new configuration file /etc/systemd/sleep.conf has been
2841 added that may be used to configure which kernel operation
2842 systemd is supposed to execute when "suspend", "hibernate"
2843 or "hybrid-sleep" is requested. This makes the new kernel
2844 "freeze" state accessible to the user.
2846 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
2847 the passed argument if applicable.
2849 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
2850 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
2851 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
2852 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
2853 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
2854 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
2855 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
2860 * The output of 'systemctl list-jobs' got some polishing. The
2861 '--type=' argument may now be passed more than once. A new
2862 command 'systemctl list-sockets' has been added which shows
2863 a list of kernel sockets systemd is listening on with the
2864 socket units they belong to, plus the units these socket
2867 * The experimental libsystemd-bus library got substantial
2868 updates to work in conjunction with the (also experimental)
2869 kdbus kernel project. It works well enough to exchange
2870 messages with some sophistication. Note that kdbus is not
2871 ready yet, and the library is mostly an elaborate test case
2872 for now, and not installable.
2874 * systemd gained a new unit 'systemd-static-nodes.service'
2875 that generates static device nodes earlier during boot, and
2876 can run in conjunction with udev.
2878 * libsystemd-login gained a new call sd_pid_get_user_unit()
2879 to retrieve the user systemd unit a process is running
2880 in. This is useful for systems where systemd is used as
2883 * systemd-nspawn now places all containers in the new /machine
2884 top-level cgroup directory in the name=systemd
2885 hierarchy. libvirt will soon do the same, so that we get a
2886 uniform separation of /system, /user and /machine for system
2887 services, user processes and containers/virtual
2888 machines. This new cgroup hierarchy is also useful to stick
2889 stable names to specific container instances, which can be
2890 recognized later this way (this name may be controlled
2891 via systemd-nspawn's new -M switch). libsystemd-login also
2892 gained a new call sd_pid_get_machine_name() to retrieve the
2893 name of the container/VM a specific process belongs to.
2895 * bootchart can now store its data in the journal.
2897 * libsystemd-journal gained a new call
2898 sd_journal_add_conjunction() for AND expressions to the
2899 matching logic. This can be used to express more complex
2900 logical expressions.
2902 * journactl can now take multiple --unit= and --user-unit=
2905 * The cryptsetup logic now understands the "luks.key=" kernel
2906 command line switch for specifying a file to read the
2907 decryption key from. Also, if a configured key file is not
2908 found the tool will now automatically fall back to prompting
2911 * Python systemd.journal module was updated to wrap recently
2912 added functions from libsystemd-journal. The interface was
2913 changed to bring the low level interface in s.j._Reader
2914 closer to the C API, and the high level interface in
2915 s.j.Reader was updated to wrap and convert all data about
2918 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
2919 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
2920 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
2921 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
2922 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
2923 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2927 * journalctl --update-catalog now understands a new --root=
2928 option to operate on catalogs found in a different root
2931 * During shutdown after systemd has terminated all running
2932 services a final killing loop kills all remaining left-over
2933 processes. We will now print the name of these processes
2934 when we send SIGKILL to them, since this usually indicates a
2937 * If /etc/crypttab refers to password files stored on
2938 configured mount points automatic dependencies will now be
2939 generated to ensure the specific mount is established first
2940 before the key file is attempted to be read.
2942 * 'systemctl status' will now show information about the
2943 network sockets a socket unit is listening on.
2945 * 'systemctl status' will also shown information about any
2946 drop-in configuration file for units. (Drop-In configuration
2947 files in this context are files such as
2948 /etc/systemd/systemd/foobar.service.d/*.conf)
2950 * systemd-cgtop now optionally shows summed up CPU times of
2951 cgroups. Press '%' while running cgtop to switch between
2952 percentage and absolute mode. This is useful to determine
2953 which cgroups use up the most CPU time over the entire
2954 runtime of the system. systemd-cgtop has also been updated
2955 to be 'pipeable' for processing with further shell tools.
2957 * 'hostnamectl set-hostname' will now allow setting of FQDN
2960 * The formatting and parsing of time span values has been
2961 changed. The parser now understands fractional expressions
2962 such as "5.5h". The formatter will now output fractional
2963 expressions for all time spans under 1min, i.e. "5.123456s"
2964 rather than "5s 123ms 456us". For time spans under 1s
2965 millisecond values are shown, for those under 1ms
2966 microsecond values are shown. This should greatly improve
2967 all time-related output of systemd.
2969 * libsystemd-login and libsystemd-journal gained new
2970 functions for querying the poll() events mask and poll()
2971 timeout value for integration into arbitrary event
2974 * localectl gained the ability to list available X11 keymaps
2975 (models, layouts, variants, options).
2977 * 'systemd-analyze dot' gained the ability to filter for
2978 specific units via shell-style globs, to create smaller,
2979 more useful graphs. I.e. it is now possible to create simple
2980 graphs of all the dependencies between only target units, or
2981 of all units that Avahi has dependencies with.
2983 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2984 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2985 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2986 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2987 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2988 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2989 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2993 * The boot-time readahead implementation for rotating media
2994 will now read the read-ahead data in multiple passes which
2995 consist of all read requests made in equidistant time
2996 intervals. This means instead of strictly reading read-ahead
2997 data in its physical order on disk we now try to find a
2998 middle ground between physical and access time order.
3000 * /etc/os-release files gained a new BUILD_ID= field for usage
3001 on operating systems that provide continuous builds of OS
3004 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
3005 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
3006 William Douglas, Zbigniew Jędrzejewski-Szmek
3010 * systemd-python gained an API exposing libsystemd-daemon.
3012 * The SMACK setup logic gained support for uploading CIPSO
3015 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
3016 ReadOnlyDirectories= and InaccessibleDirectories= has
3017 changed. The private /tmp and /var/tmp directories are now
3018 shared by all processes of a service (which means
3019 ExecStartPre= may now leave data in /tmp that ExecStart= of
3020 the same service can still access). When a service is
3021 stopped its temporary directories are immediately deleted
3022 (normal clean-up with tmpfiles is still done in addition to
3025 * By default, systemd will now set a couple of sysctl
3026 variables in the kernel: the safe sysrq options are turned
3027 on, IP route verification is turned on, and source routing
3028 disabled. The recently added hardlink and softlink
3029 protection of the kernel is turned on. These settings should
3030 be reasonably safe, and good defaults for all new systems.
3032 * The predictable network naming logic may now be turned off
3033 with a new kernel command line switch: net.ifnames=0.
3035 * A new libsystemd-bus module has been added that implements a
3036 pretty complete D-Bus client library. For details see:
3038 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
3040 * journald will now explicitly flush the journal files to disk
3041 at the latest 5min after each write. The file will then also
3042 be marked offline until the next write. This should increase
3043 reliability in case of a crash. The synchronization delay
3044 can be configured via SyncIntervalSec= in journald.conf.
3046 * There's a new remote-fs-setup.target unit that can be used
3047 to pull in specific services when at least one remote file
3048 system is to be mounted.
3050 * There are new targets timers.target and paths.target as
3051 canonical targets to pull user timer and path units in
3052 from. This complements sockets.target with a similar
3053 purpose for socket units.
3055 * libudev gained a new call udev_device_set_attribute_value()
3056 to set sysfs attributes of a device.
3058 * The udev daemon now sets the default number of worker
3059 processes executed in parallel based on the number of available
3060 CPUs instead of the amount of available RAM. This is supposed
3061 to provide a more reliable default and limit a too aggressive
3062 paralellism for setups with 1000s of devices connected.
3064 Contributions from: Auke Kok, Colin Walters, Cristian
3065 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
3066 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
3067 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
3068 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
3069 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
3070 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
3071 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
3072 Zbigniew Jędrzejewski-Szmek
3076 * Configuration of unit files may now be extended via drop-in
3077 files without having to edit/override the unit files
3078 themselves. More specifically, if the administrator wants to
3079 change one value for a service file foobar.service he can
3080 now do so by dropping in a configuration snippet into
3081 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
3082 will load all these snippets and apply them on top of the
3083 main unit configuration file, possibly extending or
3084 overriding its settings. Using these drop-in snippets is
3085 generally nicer than the two earlier options for changing
3086 unit files locally: copying the files from
3087 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
3088 them there; or creating a new file in /etc/systemd/system/
3089 that incorporates the original one via ".include". Drop-in
3090 snippets into these .d/ directories can be placed in any
3091 directory systemd looks for units in, and the usual
3092 overriding semantics between /usr/lib, /etc and /run apply
3095 * Most unit file settings which take lists of items can now be
3096 reset by assigning the empty string to them. For example,
3097 normally, settings such as Environment=FOO=BAR append a new
3098 environment variable assignment to the environment block,
3099 each time they are used. By assigning Environment= the empty
3100 string the environment block can be reset to empty. This is
3101 particularly useful with the .d/*.conf drop-in snippets
3102 mentioned above, since this adds the ability to reset list
3103 settings from vendor unit files via these drop-ins.
3105 * systemctl gained a new "list-dependencies" command for
3106 listing the dependencies of a unit recursively.
3108 * Inhibitors are now honored and listed by "systemctl
3109 suspend", "systemctl poweroff" (and similar) too, not only
3110 GNOME. These commands will also list active sessions by
3113 * Resource limits (as exposed by the various control group
3114 controllers) can now be controlled dynamically at runtime
3115 for all units. More specifically, you can now use a command
3116 like "systemctl set-cgroup-attr foobar.service cpu.shares
3117 2000" to alter the CPU shares a specific service gets. These
3118 settings are stored persistently on disk, and thus allow the
3119 administrator to easily adjust the resource usage of
3120 services with a few simple commands. This dynamic resource
3121 management logic is also available to other programs via the
3122 bus. Almost any kernel cgroup attribute and controller is
3125 * systemd-vconsole-setup will now copy all font settings to
3126 all allocated VTs, where it previously applied them only to
3129 * libsystemd-login gained the new sd_session_get_tty() API
3132 * This release drops support for a few legacy or
3133 distribution-specific LSB facility names when parsing init
3134 scripts: $x-display-manager, $mail-transfer-agent,
3135 $mail-transport-agent, $mail-transfer-agent, $smtp,
3136 $null. Also, the mail-transfer-agent.target unit backing
3137 this has been removed. Distributions which want to retain
3138 compatibility with this should carry the burden for
3139 supporting this themselves and patch support for these back
3140 in, if they really need to. Also, the facilities $syslog and
3141 $local_fs are now ignored, since systemd does not support
3142 early-boot LSB init scripts anymore, and these facilities
3143 are implied anyway for normal services. syslog.target has
3146 * There are new bus calls on PID1's Manager object for
3147 cancelling jobs, and removing snapshot units. Previously,
3148 both calls were only available on the Job and Snapshot
3151 * systemd-journal-gatewayd gained SSL support.
3153 * The various "environment" files, such as /etc/locale.conf
3154 now support continuation lines with a backslash ("\") as
3155 last character in the line, similarly in style (but different)
3156 to how this is supported in shells.
3158 * For normal user processes the _SYSTEMD_USER_UNIT= field is
3159 now implicitly appended to every log entry logged. systemctl
3160 has been updated to filter by this field when operating on a
3161 user systemd instance.
3163 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
3164 CAP_AUDIT_CONTROL capabilities to the capabilities set for
3165 the container. This makes it easier to boot unmodified
3166 Fedora systems in a container, which however still requires
3167 audit=0 to be passed on the kernel command line. Auditing in
3168 kernel and userspace is unfortunately still too broken in
3169 context of containers, hence we recommend compiling it out
3170 of the kernel or using audit=0. Hopefully this will be fixed
3171 one day for good in the kernel.
3173 * nspawn gained the new --bind= and --bind-ro= parameters to
3174 bind mount specific directories from the host into the
3177 * nspawn will now mount its own devpts file system instance
3178 into the container, in order not to leak pty devices from
3179 the host into the container.
3181 * systemd will now read the firmware boot time performance
3182 information from the EFI variables, if the used boot loader
3183 supports this, and takes it into account for boot performance
3184 analysis via "systemd-analyze". This is currently supported
3185 only in conjunction with Gummiboot, but could be supported
3186 by other boot loaders too. For details see:
3188 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
3190 * A new generator has been added that automatically mounts the
3191 EFI System Partition (ESP) to /boot, if that directory
3192 exists, is empty, and no other file system has been
3193 configured to be mounted there.
3195 * logind will now send out PrepareForSleep(false) out
3196 unconditionally, after coming back from suspend. This may be
3197 used by applications as asynchronous notification for
3198 system resume events.
3200 * "systemctl unlock-sessions" has been added, that allows
3201 unlocking the screens of all user sessions at once, similar
3202 to how "systemctl lock-sessions" already locked all users
3203 sessions. This is backed by a new D-Bus call UnlockSessions().
3205 * "loginctl seat-status" will now show the master device of a
3206 seat. (i.e. the device of a seat that needs to be around for
3207 the seat to be considered available, usually the graphics
3210 * tmpfiles gained a new "X" line type, that allows
3211 configuration of files and directories (with wildcards) that
3212 shall be excluded from automatic cleanup ("aging").
3214 * udev default rules set the device node permissions now only
3215 at "add" events, and do not change them any longer with a
3216 later "change" event.
3218 * The log messages for lid events and power/sleep keypresses
3219 now carry a message ID.
3221 * We now have a substantially larger unit test suite, but this
3222 continues to be work in progress.
3224 * udevadm hwdb gained a new --root= parameter to change the
3225 root directory to operate relative to.
3227 * logind will now issue a background sync() request to the kernel
3228 early at shutdown, so that dirty buffers are flushed to disk early
3229 instead of at the last moment, in order to optimize shutdown
3232 * A new bootctl tool has been added that is an interface for
3233 certain boot loader operations. This is currently a preview
3234 and is likely to be extended into a small mechanism daemon
3235 like timedated, localed, hostnamed, and can be used by
3236 graphical UIs to enumerate available boot options, and
3237 request boot into firmware operations.
3239 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
3240 the rest of the package. It also has been updated to work
3241 correctly in initrds.
3243 * Policykit previously has been runtime optional, and is now
3244 also compile time optional via a configure switch.
3246 * systemd-analyze has been reimplemented in C. Also "systemctl
3247 dot" has moved into systemd-analyze.
3249 * "systemctl status" with no further parameters will now print
3250 the status of all active or failed units.
3252 * Operations such as "systemctl start" can now be executed
3253 with a new mode "--irreversible" which may be used to queue
3254 operations that cannot accidentally be reversed by a later
3255 job queuing. This is by default used to make shutdown
3256 requests more robust.
3258 * The Python API of systemd now gained a new module for
3259 reading journal files.
3261 * A new tool kernel-install has been added that can install
3262 kernel images according to the Boot Loader Specification:
3264 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
3266 * Boot time console output has been improved to provide
3267 animated boot time output for hanging jobs.
3269 * A new tool systemd-activate has been added which can be used
3270 to test socket activation with, directly from the command
3271 line. This should make it much easier to test and debug
3272 socket activation in daemons.
3274 * journalctl gained a new "--reverse" (or -r) option to show
3275 journal output in reverse order (i.e. newest line first).
3277 * journalctl gained a new "--pager-end" (or -e) option to jump
3278 to immediately jump to the end of the journal in the
3279 pager. This is only supported in conjunction with "less".
3281 * journalctl gained a new "--user-unit=" option, that works
3282 similarly to "--unit=" but filters for user units rather than
3285 * A number of unit files to ease adoption of systemd in
3286 initrds has been added. This moves some minimal logic from
3287 the various initrd implementations into systemd proper.
3289 * The journal files are now owned by a new group
3290 "systemd-journal", which exists specifically to allow access
3291 to the journal, and nothing else. Previously, we used the
3292 "adm" group for that, which however possibly covers more
3293 than just journal/log file access. This new group is now
3294 already used by systemd-journal-gatewayd to ensure this
3295 daemon gets access to the journal files and as little else
3296 as possible. Note that "make install" will also set FS ACLs
3297 up for /var/log/journal to give "adm" and "wheel" read
3298 access to it, in addition to "systemd-journal" which owns
3299 the journal files. We recommend that packaging scripts also
3300 add read access to "adm" + "wheel" to /var/log/journal, and
3301 all existing/future journal files. To normal users and
3302 administrators little changes, however packagers need to
3303 ensure to create the "systemd-journal" system group at
3304 package installation time.
3306 * The systemd-journal-gatewayd now runs as unprivileged user
3307 systemd-journal-gateway:systemd-journal-gateway. Packaging
3308 scripts need to create these system user/group at
3311 * timedated now exposes a new boolean property CanNTP that
3312 indicates whether a local NTP service is available or not.
3314 * systemd-detect-virt will now also detect xen PVs
3316 * The pstore file system is now mounted by default, if it is
3319 * In addition to the SELinux and IMA policies we will now also
3320 load SMACK policies at early boot.
3322 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
3323 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
3324 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
3325 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
3326 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
3327 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
3328 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
3329 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
3330 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
3331 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
3332 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
3333 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
3334 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
3335 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
3339 * Timer units now support calendar time events in addition to
3340 monotonic time events. That means you can now trigger a unit
3341 based on a calendar time specification such as "Thu,Fri
3342 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
3343 or fifth day of any month of the year 2013, given that it is
3344 a thursday or friday. This brings timer event support
3345 considerably closer to cron's capabilities. For details on
3346 the supported calendar time specification language see
3349 * udev now supports a number of different naming policies for
3350 network interfaces for predictable names, and a combination
3351 of these policies is now the default. Please see this wiki
3352 document for details:
3354 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
3356 * Auke Kok's bootchart implementation has been added to the
3357 systemd tree. It is an optional component that can graph the
3358 boot in quite some detail. It is one of the best bootchart
3359 implementations around and minimal in its code and
3362 * nss-myhostname has been integrated into the systemd source
3363 tree. nss-myhostname guarantees that the local hostname
3364 always stays resolvable via NSS. It has been a weak
3365 requirement of systemd-hostnamed since a long time, and
3366 since its code is actually trivial we decided to just
3367 include it in systemd's source tree. It can be turned off
3368 with a configure switch.
3370 * The read-ahead logic is now capable of properly detecting
3371 whether a btrfs file system is on SSD or rotating media, in
3372 order to optimize the read-ahead scheme. Previously, it was
3373 only capable of detecting this on traditional file systems
3376 * In udev, additional device properties are now read from the
3377 IAB in addition to the OUI database. Also, Bluetooth company
3378 identities are attached to the devices as well.
3380 * In service files %U may be used as specifier that is
3381 replaced by the configured user name of the service.
3383 * nspawn may now be invoked without a controlling TTY. This
3384 makes it suitable for invocation as its own service. This
3385 may be used to set up a simple containerized server system
3386 using only core OS tools.
3388 * systemd and nspawn can now accept socket file descriptors
3389 when they are started for socket activation. This enables
3390 implementation of socket activated nspawn
3391 containers. i.e. think about autospawning an entire OS image
3392 when the first SSH or HTTP connection is received. We expect
3393 that similar functionality will also be added to libvirt-lxc
3396 * journalctl will now suppress ANSI color codes when
3397 presenting log data.
3399 * systemctl will no longer show control group information for
3400 a unit if a the control group is empty anyway.
3402 * logind can now automatically suspend/hibernate/shutdown the
3405 * /etc/machine-info and hostnamed now also expose the chassis
3406 type of the system. This can be used to determine whether
3407 the local system is a laptop, desktop, handset or
3408 tablet. This information may either be configured by the
3409 user/vendor or is automatically determined from ACPI and DMI
3410 information if possible.
3412 * A number of PolicyKit actions are now bound together with
3413 "imply" rules. This should simplify creating UIs because
3414 many actions will now authenticate similar ones as well.
3416 * Unit files learnt a new condition ConditionACPower= which
3417 may be used to conditionalize a unit depending on whether an
3418 AC power source is connected or not, of whether the system
3419 is running on battery power.
3421 * systemctl gained a new "is-failed" verb that may be used in
3422 shell scripts and suchlike to check whether a specific unit
3423 is in the "failed" state.
3425 * The EnvironmentFile= setting in unit files now supports file
3426 globbing, and can hence be used to easily read a number of
3427 environment files at once.
3429 * systemd will no longer detect and recognize specific
3430 distributions. All distribution-specific #ifdeffery has been
3431 removed, systemd is now fully generic and
3432 distribution-agnostic. Effectively, not too much is lost as
3433 a lot of the code is still accessible via explicit configure
3434 switches. However, support for some distribution specific
3435 legacy configuration file formats has been dropped. We
3436 recommend distributions to simply adopt the configuration
3437 files everybody else uses now and convert the old
3438 configuration from packaging scripts. Most distributions
3439 already did that. If that's not possible or desirable,
3440 distributions are welcome to forward port the specific
3441 pieces of code locally from the git history.
3443 * When logging a message about a unit systemd will now always
3444 log the unit name in the message meta data.
3446 * localectl will now also discover system locale data that is
3447 not stored in locale archives, but directly unpacked.
3449 * logind will no longer unconditionally use framebuffer
3450 devices as seat masters, i.e. as devices that are required
3451 to be existing before a seat is considered preset. Instead,
3452 it will now look for all devices that are tagged as
3453 "seat-master" in udev. By default framebuffer devices will
3454 be marked as such, but depending on local systems other
3455 devices might be marked as well. This may be used to
3456 integrate graphics cards using closed source drivers (such
3457 as NVidia ones) more nicely into logind. Note however, that
3458 we recommend using the open source NVidia drivers instead,
3459 and no udev rules for the closed-source drivers will be
3460 shipped from us upstream.
3462 Contributions from: Adam Williamson, Alessandro Crismani, Auke
3463 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
3464 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
3465 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
3466 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
3467 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
3468 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
3469 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
3470 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
3471 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
3472 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
3473 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
3478 * udev gained support for loading additional device properties
3479 from an indexed database that is keyed by vendor/product IDs
3480 and similar device identifiers. For the beginning this
3481 "hwdb" is populated with data from the well-known PCI and
3482 USB database, but also includes PNP, ACPI and OID data. In
3483 the longer run this indexed database shall grow into
3484 becoming the one central database for non-essential
3485 userspace device metadata. Previously, data from the PCI/USB
3486 database was only attached to select devices, since the
3487 lookup was a relatively expensive operation due to O(n) time
3488 complexity (with n being the number of entries in the
3489 database). Since this is now O(1), we decided to add in this
3490 data for all devices where this is available, by
3491 default. Note that the indexed database needs to be rebuilt
3492 when new data files are installed. To achieve this you need
3493 to update your packaging scripts to invoke "udevadm hwdb
3494 --update" after installation of hwdb data files. For
3495 RPM-based distributions we introduced the new
3496 %udev_hwdb_update macro for this purpose.
3498 * The Journal gained support for the "Message Catalog", an
3499 indexed database to link up additional information with
3500 journal entries. For further details please check:
3502 http://www.freedesktop.org/wiki/Software/systemd/catalog
3504 The indexed message catalog database also needs to be
3505 rebuilt after installation of message catalog files. Use
3506 "journalctl --update-catalog" for this. For RPM-based
3507 distributions we introduced the %journal_catalog_update
3508 macro for this purpose.
3510 * The Python Journal bindings gained support for the standard
3511 Python logging framework.
3513 * The Journal API gained new functions for checking whether
3514 the underlying file system of a journal file is capable of
3515 properly reporting file change notifications, or whether
3516 applications that want to reflect journal changes "live"
3517 need to recheck journal files continuously in appropriate
3520 * It is now possible to set the "age" field for tmpfiles
3521 entries to 0, indicating that files matching this entry
3522 shall always be removed when the directories are cleaned up.
3524 * coredumpctl gained a new "gdb" verb which invokes gdb
3525 right-away on the selected coredump.
3527 * There's now support for "hybrid sleep" on kernels that
3528 support this, in addition to "suspend" and "hibernate". Use
3529 "systemctl hybrid-sleep" to make use of this.
3531 * logind's HandleSuspendKey= setting (and related settings)
3532 now gained support for a new "lock" setting to simply
3533 request the screen lock on all local sessions, instead of
3534 actually executing a suspend or hibernation.
3536 * systemd will now mount the EFI variables file system by
3539 * Socket units now gained support for configuration of the
3540 SMACK security label.
3542 * timedatectl will now output the time of the last and next
3543 daylight saving change.
3545 * We dropped support for various legacy and distro-specific
3546 concepts, such as insserv, early-boot SysV services
3547 (i.e. those for non-standard runlevels such as 'b' or 'S')
3548 or ArchLinux /etc/rc.conf support. We recommend the
3549 distributions who still need support this to either continue
3550 to maintain the necessary patches downstream, or find a
3551 different solution. (Talk to us if you have questions!)
3553 * Various systemd components will now bypass PolicyKit checks
3554 for root and otherwise handle properly if PolicyKit is not
3555 found to be around. This should fix most issues for
3556 PolicyKit-less systems. Quite frankly this should have been
3557 this way since day one. It is absolutely our intention to
3558 make systemd work fine on PolicyKit-less systems, and we
3559 consider it a bug if something does not work as it should if
3560 PolicyKit is not around.
3562 * For embedded systems it is now possible to build udev and
3563 systemd without blkid and/or kmod support.
3565 * "systemctl switch-root" is now capable of switching root
3566 more than once. I.e. in addition to transitions from the
3567 initrd to the host OS it is now possible to transition to
3568 further OS images from the host. This is useful to implement
3569 offline updating tools.
3571 * Various other additions have been made to the RPM macros
3572 shipped with systemd. Use %udev_rules_update() after
3573 installing new udev rules files. %_udevhwdbdir,
3574 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
3575 %_sysctldir are now available which resolve to the right
3576 directories for packages to place various data files in.
3578 * journalctl gained the new --full switch (in addition to
3579 --all, to disable ellipsation for long messages.
3581 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
3582 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
3583 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
3584 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
3585 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
3586 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
3587 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
3588 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
3589 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
3593 * journalctl gained new --since= and --until= switches to
3594 filter by time. It also now supports nice filtering for
3595 units via --unit=/-u.
3597 * Type=oneshot services may use ExecReload= and do the
3600 * The journal daemon now supports time-based rotation and
3601 vacuuming, in addition to the usual disk-space based
3604 * The journal will now index the available field values for
3605 each field name. This enables clients to show pretty drop
3606 downs of available match values when filtering. The bash
3607 completion of journalctl has been updated
3608 accordingly. journalctl gained a new switch -F to list all
3609 values a certain field takes in the journal database.
3611 * More service events are now written as structured messages
3612 to the journal, and made recognizable via message IDs.
3614 * The timedated, localed and hostnamed mini-services which
3615 previously only provided support for changing time, locale
3616 and hostname settings from graphical DEs such as GNOME now
3617 also have a minimal (but very useful) text-based client
3618 utility each. This is probably the nicest way to changing
3619 these settings from the command line now, especially since
3620 it lists available options and is fully integrated with bash
3623 * There's now a new tool "systemd-coredumpctl" to list and
3624 extract coredumps from the journal.
3626 * We now install a README each in /var/log/ and
3627 /etc/rc.d/init.d explaining where the system logs and init
3628 scripts went. This hopefully should help folks who go to
3629 that dirs and look into the otherwise now empty void and
3630 scratch their heads.
3632 * When user-services are invoked (by systemd --user) the
3633 $MANAGERPID env var is set to the PID of systemd.
3635 * SIGRTMIN+24 when sent to a --user instance will now result
3636 in immediate termination of systemd.
3638 * gatewayd received numerous feature additions such as a
3639 "follow" mode, for live syncing and filtering.
3641 * browse.html now allows filtering and showing detailed
3642 information on specific entries. Keyboard navigation and
3643 mouse screen support has been added.
3645 * gatewayd/journalctl now supports HTML5/JSON
3646 Server-Sent-Events as output.
3648 * The SysV init script compatibility logic will now
3649 heuristically determine whether a script supports the
3650 "reload" verb, and only then make this available as
3653 * "systemctl status --follow" has been removed, use "journalctl
3656 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3657 have been removed since they are hardly useful to be
3660 * And I'd like to take the opportunity to specifically mention
3661 Zbigniew for his great contributions. Zbigniew, you rock!
3663 Contributions from: Andrew Eikum, Christian Hesse, Colin
3664 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3665 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3666 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3667 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3668 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3669 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3673 * If /etc/vconsole.conf is non-existent or empty we will no
3674 longer load any console font or key map at boot by
3675 default. Instead the kernel defaults will be left
3676 intact. This is definitely the right thing to do, as no
3677 configuration should mean no configuration, and hard-coding
3678 font names that are different on all archs is probably a bad
3679 idea. Also, the kernel default key map and font should be
3680 good enough for most cases anyway, and mostly identical to
3681 the userspace fonts/key maps we previously overloaded them
3682 with. If distributions want to continue to default to a
3683 non-kernel font or key map they should ship a default
3684 /etc/vconsole.conf with the appropriate contents.
3686 Contributions from: Colin Walters, Daniel J Walsh, Dave
3687 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3688 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3692 * journalctl gained a new --cursor= switch to show entries
3693 starting from the specified location in the journal.
3695 * We now enforce a size limit on journal entry fields exported
3696 with "-o json" in journalctl. Fields larger than 4K will be
3697 assigned null. This can be turned off with --all.
3699 * An (optional) journal gateway daemon is now available as
3700 "systemd-journal-gatewayd.service". This service provides
3701 access to the journal via HTTP and JSON. This functionality
3702 will be used to implement live log synchronization in both
3703 pull and push modes, but has various other users too, such
3704 as easy log access for debugging of embedded devices. Right
3705 now it is already useful to retrieve the journal via HTTP:
3707 # systemctl start systemd-journal-gatewayd.service
3708 # wget http://localhost:19531/entries
3710 This will download the journal contents in a
3711 /var/log/messages compatible format. The same as JSON:
3713 # curl -H"Accept: application/json" http://localhost:19531/entries
3715 This service is also accessible via a web browser where a
3716 single static HTML5 app is served that uses the JSON logic
3717 to enable the user to do some basic browsing of the
3718 journal. This will be extended later on. Here's an example
3719 screenshot of this app in its current state:
3721 http://0pointer.de/public/journal-gatewayd
3723 Contributions from: Kay Sievers, Lennart Poettering, Robert
3724 Milasan, Tom Gundersen
3728 * The bash completion logic is now available for journalctl
3731 * We do not mount the "cpuset" controller anymore together with
3732 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
3733 started if no parameters are assigned to it. "cpuset" hence
3734 broke code that assumed it it could create "cpu" groups and
3737 * journalctl -f will now subscribe to terminal size changes,
3738 and line break accordingly.
3740 Contributions from: Dave Reisner, Kay Sievers, Lennart
3741 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3745 * nspawn will now create a symlink /etc/localtime in the
3746 container environment, copying the host's timezone
3747 setting. Previously this has been done via a bind mount, but
3748 since symlinks cannot be bind mounted this has now been
3749 changed to create/update the appropriate symlink.
3751 * journalctl -n's line number argument is now optional, and
3752 will default to 10 if omitted.
3754 * journald will now log the maximum size the journal files may
3755 take up on disk. This is particularly useful if the default
3756 built-in logic of determining this parameter from the file
3757 system size is used. Use "systemctl status
3758 systemd-journald.service" to see this information.
3760 * The multi-seat X wrapper tool has been stripped down. As X
3761 is now capable of enumerating graphics devices via udev in a
3762 seat-aware way the wrapper is not strictly necessary
3763 anymore. A stripped down temporary stop-gap is still shipped
3764 until the upstream display managers have been updated to
3765 fully support the new X logic. Expect this wrapper to be
3766 removed entirely in one of the next releases.
3768 * HandleSleepKey= in logind.conf has been split up into
3769 HandleSuspendKey= and HandleHibernateKey=. The old setting
3770 is not available anymore. X11 and the kernel are
3771 distuingishing between these keys and we should too. This
3772 also means the inhibition lock for these keys has been split
3775 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
3776 Poettering, Lukas Nykryn, Václav Pavlín
3780 * Whenever a unit changes state we will now log this to the
3781 journal and show along the unit's own log output in
3784 * ConditionPathIsMountPoint= can now properly detect bind
3785 mount points too. (Previously, a bind mount of one file
3786 system to another place in the same file system could not be
3787 detected as mount, since they shared struct stat's st_dev
3790 * We will now mount the cgroup controllers cpu, cpuacct,
3791 cpuset and the controllers net_cls, net_prio together by
3794 * nspawn containers will now have a virtualized boot
3795 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
3796 over with a randomized ID at container initialization). This
3797 has the effect of making "journalctl -b" do the right thing
3800 * The JSON output journal serialization has been updated not
3801 to generate "endless" list objects anymore, but rather one
3802 JSON object per line. This is more in line how most JSON
3803 parsers expect JSON objects. The new output mode
3804 "json-pretty" has been added to provide similar output, but
3805 neatly aligned for readability by humans.
3807 * We dropped all explicit sync() invocations in the shutdown
3808 code. The kernel does this implicitly anyway in the kernel
3809 reboot() syscall. halt(8)'s -n option is now a compatibility
3812 * We now support virtualized reboot() in containers, as
3813 supported by newer kernels. We will fall back to exit() if
3814 CAP_SYS_REBOOT is not available to the container. Also,
3815 nspawn makes use of this now and will actually reboot the
3816 container if the containerized OS asks for that.
3818 * journalctl will only show local log output by default
3819 now. Use --merge (-m) to show remote log output, too.
3821 * libsystemd-journal gained the new sd_journal_get_usage()
3822 call to determine the current disk usage of all journal
3823 files. This is exposed in the new "journalctl --disk-usage"
3826 * journald gained a new configuration setting SplitMode= in
3827 journald.conf which may be used to control how user journals
3828 are split off. See journald.conf(5) for details.
3830 * A new condition type ConditionFileNotEmpty= has been added.
3832 * tmpfiles' "w" lines now support file globbing, to write
3833 multiple files at once.
3835 * We added Python bindings for the journal submission
3836 APIs. More Python APIs for a number of selected APIs will
3837 likely follow. Note that we intend to add native bindings
3838 only for the Python language, as we consider it common
3839 enough to deserve bindings shipped within systemd. There are
3840 various projects outside of systemd that provide bindings
3841 for languages such as PHP or Lua.
3843 * Many conditions will now resolve specifiers such as %i. In
3844 addition, PathChanged= and related directives of .path units
3845 now support specifiers as well.
3847 * There's now a new RPM macro definition for the system preset
3850 * journald will now warn if it ca not forward a message to the
3851 syslog daemon because its socket is full.
3853 * timedated will no longer write or process /etc/timezone,
3854 except on Debian. As we do not support late mounted /usr
3855 anymore /etc/localtime always being a symlink is now safe,
3856 and hence the information in /etc/timezone is not necessary
3859 * logind will now always reserve one VT for a text getty (VT6
3860 by default). Previously if more than 6 X sessions where
3861 started they took up all the VTs with auto-spawned gettys,
3862 so that no text gettys were available anymore.
3864 * udev will now automatically inform the btrfs kernel logic
3865 about btrfs RAID components showing up. This should make
3866 simple hotplug based btrfs RAID assembly work.
3868 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
3869 (but not for its children which will stay at the kernel
3870 default). This should allow setups with a lot more listening
3873 * systemd will now always pass the configured timezone to the
3874 kernel at boot. timedated will do the same when the timezone
3877 * logind's inhibition logic has been updated. By default,
3878 logind will now handle the lid switch, the power and sleep
3879 keys all the time, even in graphical sessions. If DEs want
3880 to handle these events on their own they should take the new
3881 handle-power-key, handle-sleep-key and handle-lid-switch
3882 inhibitors during their runtime. A simple way to achieve
3883 that is to invoke the DE wrapped in an invocation of:
3885 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
3887 * Access to unit operations is now checked via SELinux taking
3888 the unit file label and client process label into account.
3890 * systemd will now notify the administrator in the journal
3891 when he over-mounts a non-empty directory.
3893 * There are new specifiers that are resolved in unit files,
3894 for the host name (%H), the machine ID (%m) and the boot ID
3897 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
3898 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
3899 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
3900 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
3901 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
3902 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3903 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
3907 * Support for reading structured kernel messages from
3908 /dev/kmsg has now been added and is enabled by default.
3910 * Support for reading kernel messages from /proc/kmsg has now
3911 been removed. If you want kernel messages in the journal
3912 make sure to run a recent kernel (>= 3.5) that supports
3913 reading structured messages from /dev/kmsg (see
3914 above). /proc/kmsg is now exclusive property of classic
3915 syslog daemons again.
3917 * The libudev API gained the new
3918 udev_device_new_from_device_id() call.
3920 * The logic for file system namespace (ReadOnlyDirectory=,
3921 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
3922 require pivot_root() anymore. This means fewer temporary
3923 directories are created below /tmp for this feature.
3925 * nspawn containers will now see and receive all submounts
3926 made on the host OS below the root file system of the
3929 * Forward Secure Sealing is now supported for Journal files,
3930 which provide cryptographical sealing of journal files so
3931 that attackers cannot alter log history anymore without this
3932 being detectable. Lennart will soon post a blog story about
3933 this explaining it in more detail.
3935 * There are two new service settings RestartPreventExitStatus=
3936 and SuccessExitStatus= which allow configuration of exit
3937 status (exit code or signal) which will be excepted from the
3938 restart logic, resp. consider successful.
3940 * journalctl gained the new --verify switch that can be used
3941 to check the integrity of the structure of journal files and
3942 (if Forward Secure Sealing is enabled) the contents of
3945 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3946 and similar symlinks pre-created. This makes running shells
3947 as container init process a lot more fun.
3949 * The fstab support can now handle PARTUUID= and PARTLABEL=
3952 * A new ConditionHost= condition has been added to match
3953 against the hostname (with globs) and machine ID. This is
3954 useful for clusters where a single OS image is used to
3955 provision a large number of hosts which shall run slightly
3956 different sets of services.
3958 * Services which hit the restart limit will now be placed in a
3961 Contributions from: Bertram Poettering, Dave Reisner, Huang
3962 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3963 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3967 * When running in --user mode systemd will now become a
3968 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3969 tree a lot more organized.
3971 * A new PartOf= unit dependency type has been introduced that
3972 may be used to group services in a natural way.
3974 * "systemctl enable" may now be used to enable instances of
3977 * journalctl now prints error log levels in red, and
3978 warning/notice log levels in bright white. It also supports
3979 filtering by log level now.
3981 * cgtop gained a new -n switch (similar to top), to configure
3982 the maximum number of iterations to run for. It also gained
3983 -b, to run in batch mode (accepting no input).
3985 * The suffix ".service" may now be omitted on most systemctl
3986 command lines involving service unit names.
3988 * There's a new bus call in logind to lock all sessions, as
3989 well as a loginctl verb for it "lock-sessions".
3991 * libsystemd-logind.so gained a new call sd_journal_perror()
3992 that works similar to libc perror() but logs to the journal
3993 and encodes structured information about the error number.
3995 * /etc/crypttab entries now understand the new keyfile-size=
3998 * shutdown(8) now can send a (configurable) wall message when
3999 a shutdown is cancelled.
4001 * The mount propagation mode for the root file system will now
4002 default to "shared", which is useful to make containers work
4003 nicely out-of-the-box so that they receive new mounts from
4004 the host. This can be undone locally by running "mount
4005 --make-rprivate /" if needed.
4007 * The prefdm.service file has been removed. Distributions
4008 should maintain this unit downstream if they intend to keep
4009 it around. However, we recommend writing normal unit files
4010 for display managers instead.
4012 * Since systemd is a crucial part of the OS we will now
4013 default to a number of compiler switches that improve
4014 security (hardening) such as read-only relocations, stack
4015 protection, and suchlike.
4017 * The TimeoutSec= setting for services is now split into
4018 TimeoutStartSec= and TimeoutStopSec= to allow configuration
4019 of individual time outs for the start and the stop phase of
4022 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
4023 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
4024 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
4025 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
4026 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
4027 Gundersen, Zbigniew Jędrzejewski-Szmek
4031 * The journal and id128 C APIs are now fully documented as man
4034 * Extra safety checks have been added when transitioning from
4035 the initial RAM disk to the main system to avoid accidental
4038 * /etc/crypttab entries now understand the new keyfile-offset=
4041 * systemctl -t can now be used to filter by unit load state.
4043 * The journal C API gained the new sd_journal_wait() call to
4044 make writing synchronous journal clients easier.
4046 * journalctl gained the new -D switch to show journals from a
4049 * journalctl now displays a special marker between log
4050 messages of two different boots.
4052 * The journal is now explicitly flushed to /var via a service
4053 systemd-journal-flush.service, rather than implicitly simply
4054 by seeing /var/log/journal to be writable.
4056 * journalctl (and the journal C APIs) can now match for much
4057 more complex expressions, with alternatives and
4060 * When transitioning from the initial RAM disk to the main
4061 system we will now kill all processes in a killing spree to
4062 ensure no processes stay around by accident.
4064 * Three new specifiers may be used in unit files: %u, %h, %s
4065 resolve to the user name, user home directory resp. user
4066 shell. This is useful for running systemd user instances.
4068 * We now automatically rotate journal files if their data
4069 object hash table gets a fill level > 75%. We also size the
4070 hash table based on the configured maximum file size. This
4071 together should lower hash collisions drastically and thus
4072 speed things up a bit.
4074 * journalctl gained the new "--header" switch to introspect
4075 header data of journal files.
4077 * A new setting SystemCallFilters= has been added to services
4078 which may be used to apply blacklists or whitelists to
4079 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
4081 * nspawn gained a new --link-journal= switch (and quicker: -j)
4082 to link the container journal with the host. This makes it
4083 very easy to centralize log viewing on the host for all
4084 guests while still keeping the journal files separated.
4086 * Many bugfixes and optimizations
4088 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
4089 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
4090 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
4095 * Several tools now understand kernel command line arguments,
4096 which are only read when run in an initial RAM disk. They
4097 usually follow closely their normal counterparts, but are
4100 * There's a new tool to analyze the readahead files that are
4101 automatically generated at boot. Use:
4103 /usr/lib/systemd/systemd-readahead analyze /.readahead
4105 * We now provide an early debug shell on tty9 if this enabled. Use:
4107 systemctl enable debug-shell.service
4109 * All plymouth related units have been moved into the Plymouth
4110 package. Please make sure to upgrade your Plymouth version
4113 * systemd-tmpfiles now supports getting passed the basename of
4114 a configuration file only, in which case it will look for it
4115 in all appropriate directories automatically.
4117 * udevadm info now takes a /dev or /sys path as argument, and
4118 does the right thing. Example:
4120 udevadm info /dev/sda
4121 udevadm info /sys/class/block/sda
4123 * systemctl now prints a warning if a unit is stopped but a
4124 unit that might trigger it continues to run. Example: a
4125 service is stopped but the socket that activates it is left
4128 * "systemctl status" will now mention if the log output was
4129 shortened due to rotation since a service has been started.
4131 * The journal API now exposes functions to determine the
4132 "cutoff" times due to rotation.
4134 * journald now understands SIGUSR1 and SIGUSR2 for triggering
4135 immediately flushing of runtime logs to /var if possible,
4136 resp. for triggering immediate rotation of the journal
4139 * It is now considered an error if a service is attempted to
4140 be stopped that is not loaded.
4142 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
4144 * systemd-analyze now supports Python 3
4146 * tmpfiles now supports cleaning up directories via aging
4147 where the first level dirs are always kept around but
4148 directories beneath it automatically aged. This is enabled
4149 by prefixing the age field with '~'.
4151 * Seat objects now expose CanGraphical, CanTTY properties
4152 which is required to deal with very fast bootups where the
4153 display manager might be running before the graphics drivers
4154 completed initialization.
4156 * Seat objects now expose a State property.
4158 * We now include RPM macros for service enabling/disabling
4159 based on the preset logic. We recommend RPM based
4160 distributions to make use of these macros if possible. This
4161 makes it simpler to reuse RPM spec files across
4164 * We now make sure that the collected systemd unit name is
4165 always valid when services log to the journal via
4168 * There's a new man page kernel-command-line(7) detailing all
4169 command line options we understand.
4171 * The fstab generator may now be disabled at boot by passing
4172 fstab=0 on the kernel command line.
4174 * A new kernel command line option modules-load= is now understood
4175 to load a specific kernel module statically, early at boot.
4177 * Unit names specified on the systemctl command line are now
4178 automatically escaped as needed. Also, if file system or
4179 device paths are specified they are automatically turned
4180 into the appropriate mount or device unit names. Example:
4182 systemctl status /home
4183 systemctl status /dev/sda
4185 * The SysVConsole= configuration option has been removed from
4186 system.conf parsing.
4188 * The SysV search path is no longer exported on the D-Bus
4191 * The Names= option is been removed from unit file parsing.
4193 * There's a new man page bootup(7) detailing the boot process.
4195 * Every unit and every generator we ship with systemd now
4196 comes with full documentation. The self-explanatory boot is
4199 * A couple of services gained "systemd-" prefixes in their
4200 name if they wrap systemd code, rather than only external
4201 code. Among them fsck@.service which is now
4202 systemd-fsck@.service.
4204 * The HaveWatchdog property has been removed from the D-Bus
4207 * systemd.confirm_spawn= on the kernel command line should now
4210 * There's a new man page crypttab(5) which details all options
4211 we actually understand.
4213 * systemd-nspawn gained a new --capability= switch to pass
4214 additional capabilities to the container.
4216 * timedated will now read known NTP implementation unit names
4217 from /usr/lib/systemd/ntp-units.d/*.list,
4218 systemd-timedated-ntp.target has been removed.
4220 * journalctl gained a new switch "-b" that lists log data of
4221 the current boot only.
4223 * The notify socket is in the abstract namespace again, in
4224 order to support daemons which chroot() at start-up.
4226 * There is a new Storage= configuration option for journald
4227 which allows configuration of where log data should go. This
4228 also provides a way to disable journal logging entirely, so
4229 that data collected is only forwarded to the console, the
4230 kernel log buffer or another syslog implementation.
4232 * Many bugfixes and optimizations
4234 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
4235 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
4236 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
4237 Shawn Landden, Tom Gundersen
4241 * "systemctl help <unit>" now shows the man page if one is
4244 * Several new man pages have been added.
4246 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
4247 MaxLevelConsole= can now be specified in
4248 journald.conf. These options allow reducing the amount of
4249 data stored on disk or forwarded by the log level.
4251 * TimerSlackNSec= can now be specified in system.conf for
4252 PID1. This allows system-wide power savings.
4254 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
4255 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
4260 * logind is now capable of (optionally) handling power and
4261 sleep keys as well as the lid switch.
4263 * journalctl now understands the syntax "journalctl
4264 /usr/bin/avahi-daemon" to get all log output of a specific
4267 * CapabilityBoundingSet= in system.conf now also influences
4268 the capability bound set of usermode helpers of the kernel.
4270 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
4271 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
4272 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
4273 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
4277 * Note that we skipped 139 releases here in order to set the
4278 new version to something that is greater than both udev's
4279 and systemd's most recent version number.
4281 * udev: all udev sources are merged into the systemd source tree now.
4282 All future udev development will happen in the systemd tree. It
4283 is still fully supported to use the udev daemon and tools without
4284 systemd running, like in initramfs or other init systems. Building
4285 udev though, will require the *build* of the systemd tree, but
4286 udev can be properly *run* without systemd.
4288 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
4289 should be used to create dead device nodes as workarounds for broken
4292 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
4293 no longer supported. udev_monitor_new_from_netlink() needs to be
4294 used to subscribe to events.
4296 * udev: when udevd is started by systemd, processes which are left
4297 behind by forking them off of udev rules, are unconditionally cleaned
4298 up and killed now after the event handling has finished. Services or
4299 daemons must be started as systemd services. Services can be
4300 pulled-in by udev to get started, but they can no longer be directly
4301 forked by udev rules.
4303 * udev: the daemon binary is called systemd-udevd now and installed
4304 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
4305 to adapt to that, create symlink, or rename the binary after building
4308 * libudev no longer provides these symbols:
4309 udev_monitor_from_socket()
4310 udev_queue_get_failed_list_entry()
4311 udev_get_{dev,sys,run}_path()
4312 The versions number was bumped and symbol versioning introduced.
4314 * systemd-loginctl and systemd-journalctl have been renamed
4315 to loginctl and journalctl to match systemctl.
4317 * The config files: /etc/systemd/systemd-logind.conf and
4318 /etc/systemd/systemd-journald.conf have been renamed to
4319 logind.conf and journald.conf. Package updates should rename
4320 the files to the new names on upgrade.
4322 * For almost all files the license is now LGPL2.1+, changed
4323 from the previous GPL2.0+. Exceptions are some minor stuff
4324 of udev (which will be changed to LGPL2.1 eventually, too),
4325 and the MIT licensed sd-daemon.[ch] library that is suitable
4326 to be used as drop-in files.
4328 * systemd and logind now handle system sleep states, in
4329 particular suspending and hibernating.
4331 * logind now implements a sleep/shutdown/idle inhibiting logic
4332 suitable for a variety of uses. Soonishly Lennart will blog
4333 about this in more detail.
4335 * var-run.mount and var-lock.mount are no longer provided
4336 (which prevously bind mounted these directories to their new
4337 places). Distributions which have not converted these
4338 directories to symlinks should consider stealing these files
4339 from git history and add them downstream.
4341 * We introduced the Documentation= field for units and added
4342 this to all our shipped units. This is useful to make it
4343 easier to explore the boot and the purpose of the various
4346 * All smaller setup units (such as
4347 systemd-vconsole-setup.service) now detect properly if they
4348 are run in a container and are skipped when
4349 appropriate. This guarantees an entirely noise-free boot in
4350 Linux container environments such as systemd-nspawn.
4352 * A framework for implementing offline system updates is now
4353 integrated, for details see:
4354 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
4356 * A new service type Type=idle is available now which helps us
4357 avoiding ugly interleaving of getty output and boot status
4360 * There's now a system-wide CapabilityBoundingSet= option to
4361 globally reduce the set of capabilities for the
4362 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
4363 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
4364 even CAP_NET_ADMIN system-wide for secure systems.
4366 * There are now system-wide DefaultLimitXXX= options to
4367 globally change the defaults of the various resource limits
4368 for all units started by PID 1.
4370 * Harald Hoyer's systemd test suite has been integrated into
4371 systemd which allows easy testing of systemd builds in qemu
4372 and nspawn. (This is really awesome! Ask us for details!)
4374 * The fstab parser is now implemented as generator, not inside
4377 * systemctl will now warn you if .mount units generated from
4378 /etc/fstab are out of date due to changes in fstab that
4379 have not been read by systemd yet.
4381 * systemd is now suitable for usage in initrds. Dracut has
4382 already been updated to make use of this. With this in place
4383 initrds get a slight bit faster but primarily are much
4384 easier to introspect and debug since "systemctl status" in
4385 the host system can be used to introspect initrd services,
4386 and the journal from the initrd is kept around too.
4388 * systemd-delta has been added, a tool to explore differences
4389 between user/admin configuration and vendor defaults.
4391 * PrivateTmp= now affects both /tmp and /var/tmp.
4393 * Boot time status messages are now much prettier and feature
4394 proper english language. Booting up systemd has never been
4397 * Read-ahead pack files now include the inode number of all
4398 files to pre-cache. When the inode changes the pre-caching
4399 is not attempted. This should be nicer to deal with updated
4400 packages which might result in changes of read-ahead
4403 * We now temporaritly lower the kernel's read_ahead_kb variable
4404 when collecting read-ahead data to ensure the kernel's
4405 built-in read-ahead does not add noise to our measurements
4406 of necessary blocks to pre-cache.
4408 * There's now RequiresMountsFor= to add automatic dependencies
4409 for all mounts necessary for a specific file system path.
4411 * MountAuto= and SwapAuto= have been removed from
4412 system.conf. Mounting file systems at boot has to take place
4415 * nspawn now learned a new switch --uuid= to set the machine
4416 ID on the command line.
4418 * nspawn now learned the -b switch to automatically search
4421 * vt102 is now the default TERM for serial TTYs, upgraded from
4424 * systemd-logind now works on VT-less systems.
4426 * The build tree has been reorganized. The individual
4427 components now have directories of their own.
4429 * A new condition type ConditionPathIsReadWrite= is now available.
4431 * nspawn learned the new -C switch to create cgroups for the
4432 container in other hierarchies.
4434 * We now have support for hardware watchdogs, configurable in
4437 * The scheduled shutdown logic now has a public API.
4439 * We now mount /tmp as tmpfs by default, but this can be
4440 masked and /etc/fstab can override it.
4442 * Since udisks does not make use of /media anymore we are not
4443 mounting a tmpfs on it anymore.
4445 * journalctl gained a new --local switch to only interleave
4446 locally generated journal files.
4448 * We can now load the IMA policy at boot automatically.
4450 * The GTK tools have been split off into a systemd-ui.
4452 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
4453 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
4454 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
4455 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
4456 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
4457 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
4458 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
4459 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
4460 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
4465 * This is mostly a bugfix release
4467 * Support optional initialization of the machine ID from the
4468 KVM or container configured UUID.
4470 * Support immediate reboots with "systemctl reboot -ff"
4472 * Show /etc/os-release data in systemd-analyze output
4474 * Many bugfixes for the journal, including endianness fixes and
4475 ensuring that disk space enforcement works
4477 * sd-login.h is C++ comptaible again
4479 * Extend the /etc/os-release format on request of the Debian
4482 * We now refuse non-UTF8 strings used in various configuration
4483 and unit files. This is done to ensure we do not pass invalid
4484 data over D-Bus or expose it elsewhere.
4486 * Register Mimo USB Screens as suitable for automatic seat
4489 * Read SELinux client context from journal clients in a race
4492 * Reorder configuration file lookup order. /etc now always
4493 overrides /run in order to allow the administrator to always
4494 and unconditionally override vendor supplied or
4495 automatically generated data.
4497 * The various user visible bits of the journal now have man
4498 pages. We still lack man pages for the journal API calls
4501 * We now ship all man pages in HTML format again in the
4504 Contributions from: Dave Reisner, Dirk Eibach, Frederic
4505 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
4506 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
4511 * This is mostly a bugfix release
4513 * systems lacking /etc/os-release are no longer supported.
4515 * Various functionality updates to libsystemd-login.so
4517 * Track class of PAM logins to distuingish greeters from
4520 Contributions from: Kay Sievers, Lennart Poettering, Michael
4525 * This is an important bugfix release for v41.
4527 * Building man pages is now optional which should be useful
4528 for those building systemd from git but unwilling to install
4531 * Watchdog support for supervising services is now usable. In
4532 a future release support for hardware watchdogs
4533 (i.e. /dev/watchdog) will be added building on this.
4535 * Service start rate limiting is now configurable and can be
4536 turned off per service. When a start rate limit is hit a
4537 reboot can automatically be triggered.
4539 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
4541 Contributions from: Benjamin Franzke, Bill Nottingham,
4542 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
4543 Schmidt, Michał Górny, Piotr Drąg
4547 * The systemd binary is installed /usr/lib/systemd/systemd now;
4548 An existing /sbin/init symlink needs to be adapted with the
4551 * The code that loads kernel modules has been ported to invoke
4552 libkmod directly, instead of modprobe. This means we do not
4553 support systems with module-init-tools anymore.
4555 * Watchdog support is now already useful, but still not
4558 * A new kernel command line option systemd.setenv= is
4559 understood to set system wide environment variables
4560 dynamically at boot.
4562 * We now limit the set of capabilities of systemd-journald.
4564 * We now set SIGPIPE to ignore by default, since it only is
4565 useful in shell pipelines, and has little use in general
4566 code. This can be disabled with IgnoreSIPIPE=no in unit
4569 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
4570 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
4575 * This is mostly a bugfix release
4577 * We now expose the reason why a service failed in the
4578 "Result" D-Bus property.
4580 * Rudimentary service watchdog support (will be completed over
4581 the next few releases.)
4583 * When systemd forks off in order execute some service we will
4584 now immediately changes its argv[0] to reflect which process
4585 it will execute. This is useful to minimize the time window
4586 with a generic argv[0], which makes bootcharts more useful
4588 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
4589 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
4590 Mike Kazantsev, Ray Strode
4594 * This is mostly a test release, but incorporates many
4597 * New systemd-cgtop tool to show control groups by their
4600 * Linking against libacl for ACLs is optional again. If
4601 disabled, support tracking device access for active logins
4602 goes becomes unavailable, and so does access to the user
4603 journals by the respective users.
4605 * If a group "adm" exists, journal files are automatically
4606 owned by them, thus allow members of this group full access
4607 to the system journal as well as all user journals.
4609 * The journal now stores the SELinux context of the logging
4610 client for all entries.
4612 * Add C++ inclusion guards to all public headers
4614 * New output mode "cat" in the journal to print only text
4615 messages, without any meta data like date or time.
4617 * Include tiny X server wrapper as a temporary stop-gap to
4618 teach XOrg udev display enumeration. This is used by display
4619 managers such as gdm, and will go away as soon as XOrg
4620 learned native udev hotplugging for display devices.
4622 * Add new systemd-cat tool for executing arbitrary programs
4623 with STDERR/STDOUT connected to the journal. Can also act as
4624 BSD logger replacement, and does so by default.
4626 * Optionally store all locally generated coredumps in the
4627 journal along with meta data.
4629 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
4630 writing short strings to files (for usage for /sys), and for
4631 creating symlinks, character and block device nodes.
4633 * New unit file option ControlGroupPersistent= to make cgroups
4634 persistent, following the mechanisms outlined in
4635 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
4637 * Support multiple local RTCs in a sane way
4639 * No longer monopolize IO when replaying readahead data on
4640 rotating disks, since we might starve non-file-system IO to
4641 death, since fanotify() will not see accesses done by blkid,
4644 * Do not show kernel threads in systemd-cgls anymore, unless
4645 requested with new -k switch.
4647 Contributions from: Dan Horák, Kay Sievers, Lennart
4648 Poettering, Michal Schmidt
4652 * This is mostly a test release, but incorporates many
4655 * The git repository moved to:
4656 git://anongit.freedesktop.org/systemd/systemd
4657 ssh://git.freedesktop.org/git/systemd/systemd
4659 * First release with the journal
4660 http://0pointer.de/blog/projects/the-journal.html
4662 * The journal replaces both systemd-kmsg-syslogd and
4663 systemd-stdout-bridge.
4665 * New sd_pid_get_unit() API call in libsystemd-logind
4667 * Many systemadm clean-ups
4669 * Introduce remote-fs-pre.target which is ordered before all
4670 remote mounts and may be used to start services before all
4673 * Added Mageia support
4675 * Add bash completion for systemd-loginctl
4677 * Actively monitor PID file creation for daemons which exit in
4678 the parent process before having finished writing the PID
4679 file in the daemon process. Daemons which do this need to be
4680 fixed (i.e. PID file creation must have finished before the
4681 parent exits), but we now react a bit more gracefully to them.
4683 * Add colourful boot output, mimicking the well-known output
4684 of existing distributions.
4686 * New option PassCredentials= for socket units, for
4687 compatibility with a recent kernel ABI breakage.
4689 * /etc/rc.local is now hooked in via a generator binary, and
4690 thus will no longer act as synchronization point during
4693 * systemctl list-unit-files now supports --root=.
4695 * systemd-tmpfiles now understands two new commands: z, Z for
4696 relabelling files according to the SELinux database. This is
4697 useful to apply SELinux labels to specific files in /sys,
4700 * Output of SysV services is now forwarded to both the console
4701 and the journal by default, not only just the console.
4703 * New man pages for all APIs from libsystemd-login.
4705 * The build tree got reorganized and a the build system is a
4706 lot more modular allowing embedded setups to specifically
4707 select the components of systemd they are interested in.
4709 * Support for Linux systems lacking the kernel VT subsystem is
4712 * configure's --with-rootdir= got renamed to
4713 --with-rootprefix= to follow the naming used by udev and
4716 * Unless specified otherwise we will now install to /usr instead
4717 of /usr/local by default.
4719 * Processes with '@' in argv[0][0] are now excluded from the
4720 final shut-down killing spree, following the logic explained
4722 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
4724 * All processes remaining in a service cgroup when we enter
4725 the START or START_PRE states are now killed with
4726 SIGKILL. That means it is no longer possible to spawn
4727 background processes from ExecStart= lines (which was never
4728 supported anyway, and bad style).
4730 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
4731 reloading of units together.
4733 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
4734 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
4735 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
4736 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
4737 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek