You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

3804 lines
181 KiB

systemd System and Service Manager
* timedated does no longer read NTP unit names from
/usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
implementations should add a:
to take over and replace systemd's NTP functionality.
* A new tool systemd-sysusers has been added. This tool
creates system users and groups in /etc/passwd and
/etc/group, based on static declarative system user/group
definitions in /usr/lib/sysusers.d/. This is useful to
enable factory resets and volatile systems that boot up with
an empty /etc directory, and thus need system users and
groups created during early boot. systemd now also ships
with two default sysusers.d/ files for the most basic
users and groups systemd and the core operating system
* A new tmpfiles snippet has been added that rebuilds the
essential files in /etc on boot, should they be missing.
* A directive for ensuring automatic clean-up of
/var/cache/man/ has been removed from the default
configuration. This line should now be shipped by the man
implementation. The necessary change has been made to the
man-db implementation. Note that you need to update your man
implementation to one that ships this line, otherwise no
automatic clean-up of /var/cache/man will take place.
* A new condition ConditionNeedsUpdate= has been added that
may conditionalize services to only run when /etc or /var
are "older" than the vendor operating system resources in
/usr. This is useful for reconstructing or updating /etc
after an offline update of /usr or a factory reset, on the
next reboot. Services that want to run once after such an
update or reset should use this condition and order
themselves before the new systemd-update-done.service, which
will mark the two directories as fully updated. A number of
service files have been added making use of this, to rebuild
the udev hardware database, the journald message catalog and
dynamic loader cache (ldconfig). The systemd-sysusers tool
described above also makes use of this now. With this in
place it is now possible to start up a minimal operating
system with /etc empty cleanly. For more information on the
concepts involved see this recent blog story:
* A new system group "input" has been introduced, and all
input device nodes get this group assigned. This is useful
for system-level software to get access to input devices. It
complements what is already done for "audio" and "video".
* systemd-networkd learnt minimal DHCPv4 server support in
addition to the existing DHCPv4 client support. It also
learnt DHCPv6 client and IPv6 Router Solicitation client
support. The DHCPv4 client gained support for static routes
passed in from the server. Note that the [DHCPv4] section
known in older systemd-networkd versions has been renamed to
[DHCP] and is now also used by the DHCPv6 client. Existing
.network files using settings of this section should be
updated, though compatibility is maintained. Optionally, the
client hostname may now be sent to the DHCP server.
* networkd gained support for vxlan virtual networks as well
as tun/tap and dummy devices.
* networkd gained support for automatic allocation of address
ranges for interfaces from a system-wide pool of
addresses. This is useful for dynamically managing a large
number of interfaces with a single network configuration
file. In particular this is useful to easily assign
appropriate IP addresses to the veth links of a large number
of nspawn instances.
* RPM macros for processing sysusers, sysctl and binfmt
drop-in snippets at package installation time have been
* The /etc/os-release file should now be placed in
/usr/lib/os-release. The old location is automatically
created as symlink. /usr/lib is the more appropriate
location of this file, since it shall actually describe the
vendor operating system shipped in /usr, and not the
configuration stored in /etc.
* .mount units gained a new boolean SloppyOptions= setting
that maps to mount(8)'s -s option which enables permissive
parsing of unknown mount options.
* tmpfiles learnt a new "L+" directive which creates a symlink
but (unlike "L") deletes a pre-existing file first, should
it already exist and not already be the correct
symlink. Similar, "b+", "c+" and "p+" directives have been
added as well, which create block and character devices, as
well as fifos in the filesystem, possibly removing any
pre-existing files of different types.
* For tmpfiles' "L", "L+", "C" and "C+" directives the final
'argument' field (which so far specified the source to
symlink/copy the files from) is now optional. If omitted the
same file os copied from /usr/share/factory/ suffixed by the
full destination path. This is useful for populating /etc
with essential files, by copying them from vendor defaults
shipped in /usr/share/factory/etc.
* A new command "systemctl preset-all" has been added that
applies the service preset settings to all installed unit
files. A new switch --preset-mode= has been added that
controls whether only enable or only disable operations
shall be executed.
* A new command "systemctl is-system-running" has been added
that allows checking the overall state of the system, for
example whether it is fully up and running.
* When the system boots up with an empty /etc, the equivalent
to "systemctl preset-all" is executed during early boot, to
make sure all default services are enabled after a factory
* systemd now contains a minimal preset file that enables the
most basic services systemd ships by default.
* Unit files' [Install] section gained a new DefaultInstance=
field for defining the default instance to create if a
template unit is enabled with no instance specified.
* A new passive target has been added
that may be used by services that need to make they run and
finish before the first LUKS cryptographic device is set up.
* The /dev/loop-control and /dev/btrfs-control device nodes
are now owned by the "disk" group by default, opening up
access to this group.
* systemd-coredump will now automatically generate a
stack trace of all core dumps taking place on the system,
based on elfutils' libdw library. This stack trace is logged
to the journal.
* systemd-coredump may now optionally store coredumps directly
on disk (in /var/lib/systemd/coredump, possibly compressed),
instead of storing them unconditionally in the journal. This
mode is the new default. A new configuration file
/etc/systemd/coredump.conf has been added to configure this
and other parameters of systemd-coredump.
* coredumpctl gained a new "info" verb to show details about a
specific coredump. A new switch "-1" has also been added
that makes sure to only show information about the most
recent entry instead of all entries. Also, as the tool is
generally useful now the "systemd-" prefix of the binary
name has been removed. Distributions that want to maintain
compatibility with the old name should add a symlink from
the old name to the new name.
* journald's SplitMode= now defaults to "uid". This makes sure
that unprivileged users can access their own coredumps with
coredumpctl without restrictions.
* New kernel command line options "systemd.wants=" (for
pulling an additional unit during boot), "systemd.mask="
(for masking a specific unit for the boot), and
"systemd.debug-shell" (for enabling the debug shell on tty9)
have been added. This is implemented in the new generator
* systemd-nspawn will now by default filter a couple of
syscalls for containers, among them those required for
kernel module loading, direct x86 IO port access, swap
management, and kexec. Most importantly though
open_by_handle_at() is now prohibited for containers,
closing a hole similar to a recently discussed vulnerability
in docker regarding access to files on file hierarchies the
container should normally not have access to. Note that for
nspawn we generally make no security claims anyway (and
this is explicitly documented in the man page), so this is
just a fix for one of the most obvious problems.
* A new man page file-hierarchy(7) has been added that
contains a minimized, modernized version of the file system
layout systemd expects, similar in style to the FHS
specification or hier(5). A new tool systemd-path(1) has
been added to query many of these paths for the local
machine and user.
* Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
longer done. Since the directory now has a per-user size
limit, and is cleaned on logout this appears unnecessary,
in particular since this now brings the lifecycle of this
directory closer in line with how IPC objects are handled.
* systemd.pc now exports a number of additional directories,
including $libdir (which is useful to identify the library
path for the primary architecture of the system), and a
couple of drop-in directories.
* udev's predictable network interface names now use the dev_port
sysfs attribute, introduced in linux 3.15 instead of dev_id to
distinguish between ports of the same PCI function. dev_id should
only be used for ports using the same HW address, hence the need
for dev_port.
* machined has been updated to export the OS version of a
container (read from /etc/os-release and
/usr/lib/os-release) on the bus. This is now shown in
"machinectl status" for a machine.
* A new service setting RestartForceExitStatus= has been
added. If configured to a set of exit signals or process
return values, the service will be restarted when the main
daemon process exits with any of them, regardless of the
Restart= setting.
* systemctl's -H switch for connecting to remote systemd
machines has been extended so that it may be used to
directly connect to a specific container on the
host. "systemctl -H root@foobar:waldi" will now connect as
user "root" to host "foobar", and then proceed directly to
the container named "waldi". Note that currently you have to
authenticate as user "root" for this to work, as entering
containers is a privileged operation.
Contributions from: Andreas Henriksson, Benjamin Steinwender,
Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
-- Berlin, 2014-07-03
* As an experimental feature, udev now tries to lock the
disk device node (flock(LOCK_SH|LOCK_NB)) while it
executes events for the disk or any of its partitions.
Applications like partitioning programs can lock the
disk device node (flock(LOCK_EX)) and claim temporary
device ownership that way; udev will entirely skip all event
handling for this disk and its partitions. If the disk
was opened for writing, the close will trigger a partition
table rescan in udev's "watch" facility, and if needed
synthesize "change" events for the disk and all its partitions.
This is now unconditionally enabled, and if it turns out to
cause major problems, we might turn it on only for specific
devices, or might need to disable it entirely. Device Mapper
devices are excluded from this logic.
* We temporarily dropped the "-l" switch for fsck invocations,
since they collide with the flock() logic above. util-linux
upstream has been changed already to avoid this conflict,
and we will readd "-l" as soon as util-linux with this
change has been released.
* The dependency on libattr has been removed. Since a long
time, the extended attribute calls have moved to glibc, and
libattr is thus unnecessary.
* Virtualization detection works without priviliges now. This
means the systemd-detect-virt binary no longer requires
CAP_SYS_PTRACE file capabilities, and our daemons can run
with fewer privileges.
* systemd-networkd now runs under its own "systemd-network"
CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
loses the ability to write to files owned by root this way.
* Similar, systemd-resolved now runs under its own
"systemd-resolve" user with no capabilities remaining.
* Similar, systemd-bus-proxyd now runs under its own
"systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
* systemd-networkd gained support for setting up "veth"
virtual ethernet devices for container connectivity, as well
as GRE and VTI tunnels.
* systemd-networkd will no longer automatically attempt to
manually load kernel modules necessary for certain tunnel
transports. Instead, it is assumed the kernel loads them
automatically when required. This only works correctly on
very new kernels. On older kernels, please consider adding
the kernel modules to /etc/modules-load.d/ as a work-around.
* The resolv.conf file systemd-resolved generates has been
moved to /run/systemd/resolve/. If you have a symlink from
/etc/resolv.conf, it might be necessary to correct it.
* Two new service settings, ProtectedHome= and ProtectedSystem=,
have been added. When enabled, they will make the user data
(such as /home) inaccessible or read-only and the system
(such as /usr) read-only, for specific services. This allows
very light-weight per-service sandboxing to avoid
modifications of user data or system files from
services. These two new switches have been enabled for all
of systemd's long-running services, where appropriate.
* Socket units gained new SocketUser= and SocketGroup=
settings to set the owner user and group of AF_UNIX sockets
and FIFOs in the file system.
* Socket units gained a new RemoveOnStop= setting. If enabled,
all FIFOS and sockets in the file system will be removed
when the specific socket unit is stopped.
* Socket units gained a new Symlinks= setting. It takes a list
of symlinks to create to file system sockets or FIFOs
created by the specific Unix sockets. This is useful to
manage symlinks to socket nodes with the same life-cycle as
the socket itself.
* The /dev/log socket and /dev/initctl FIFO have been moved to
/run, and have been replaced by symlinks. This allows
connecting to these facilities even if PrivateDevices=yes is
used for a service (which makes /dev/log itself unavailable,
but /run is left). This also has the benefit of ensuring
that /dev only contains device nodes, directories and
symlinks, and nothing else.
* sd-daemon gained two new calls sd_pid_notify() and
sd_pid_notifyf(). They are similar to sd_notify() and
sd_notifyf(), but allow overriding of the source PID of
notification messages if permissions permit this. This is
useful to send notify messages on behalf of a different
process (for example, the parent process). The
systemd-notify tool has been updated to make use of this
when sending messages (so that notification messages now
originate from the shell script invoking systemd-notify and
not the systemd-notify process itself. This should minimize
a race where systemd fails to associate notification
messages to services when the originating process already
* A new "on-abnormal" setting for Restart= has been added. If
set, it will result in automatic restarts on all "abnormal"
reasons for a process to exit, which includes unclean
signals, core dumps, timeouts and watchdog timeouts, but
does not include clean and unclean exit codes or clean
signals. Restart=on-abnormal is an alternative for
Restart=on-failure for services that shall be able to
terminate and avoid restarts on certain errors, by
indicating so with an unclean exit code. Restart=on-failure
or Restart=on-abnormal is now the recommended setting for
all long-running services.
* If the InaccessibleDirectories= service setting points to a
mount point (or if there are any submounts contained within
it), it is now attempted to completely unmount it, to make
the file systems truly unavailable for the respective
* The ReadOnlyDirectories= service setting and
systemd-nspawn's --read-only parameter are now recursively
applied to all submounts, too.
* Mount units may now be created transiently via the bus APIs.
* The support for SysV and LSB init scripts has been removed
from the systemd daemon itself. Instead, it is now
implemented as a generator that creates native systemd units
from these scripts when needed. This enables us to remove a
substantial amount of legacy code from PID 1, following the
fact that many distributions only ship a very small number
of LSB/SysV init scripts nowadays.
* Priviliged Xen (dom0) domains are not considered
virtualization anymore by the virtualization detection
logic. After all, they generally have unrestricted access to
the hardware and usually are used to manage the unprivileged
(domU) domains.
* systemd-tmpfiles gained a new "C" line type, for copying
files or entire directories.
* systemd-tmpfiles "m" lines are now fully equivalent to "z"
lines. So far, they have been non-globbing versions of the
latter, and have thus been redundant. In future, it is
recommended to only use "z". "m" has hence been removed
from the documentation, even though it stays supported.
* A tmpfiles snippet to recreate the most basic structure in
/var has been added. This is enough to create the /var/run →
/run symlink and create a couple of structural
directories. This allows systems to boot up with an empty or
volatile /var. Of course, while with this change, the core OS
now is capable with dealing with a volatile /var, not all
user services are ready for it. However, we hope that sooner
or later, many service daemons will be changed upstream so
that they are able to automatically create their necessary
directories in /var at boot, should they be missing. This is
the first step to allow state-less systems that only require
the vendor image for /usr to boot.
* systemd-nspawn has gained a new --tmpfs= switch to mount an
empty tmpfs instance to a specific directory. This is
particularly useful for making use of the automatic
reconstruction of /var (see above), by passing --tmpfs=/var.
* Access modes specified in tmpfiles snippets may now be
prefixed with "~", which indicates that they shall be masked
by whether the existing file or directly is currently
writable, readable or executable at all. Also, if specified,
the sgid/suid/sticky bits will be masked for all
* A new passive target unit "" has been
added which is useful for services that shall run before any
network is configured, for example firewall scripts.
* The "floppy" group that previously owned the /dev/fd*
devices is no longer used. The "disk" group is now used
instead. Distributions should probably deprecate usage of
this group.
Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
-- Berlin, 2014-06-11
* A new "systemd-timesyncd" daemon has been added for
synchronizing the system clock across the network. It
implements an SNTP client. In contrast to NTP
implementations such as chrony or the NTP reference server,
this only implements a client side, and does not bother with
the full NTP complexity, focusing only on querying time from
one remote server and synchronizing the local clock to
it. Unless you intend to serve NTP to networked clients or
want to connect to local hardware clocks, this simple NTP
client should be more than appropriate for most
installations. The daemon runs with minimal privileges, and
has been hooked up with networkd to only operate when
network connectivity is available. The daemon saves the
current clock to disk every time a new NTP sync has been
acquired, and uses this to possibly correct the system clock
early at bootup, in order to accommodate for systems that
lack an RTC such as the Raspberry Pi and embedded devices,
and to make sure that time monotonically progresses on these
systems, even if it is not always correct. To make use of
this daemon, a new system user and group "systemd-timesync"
needs to be created on installation of systemd.
* The queue "seqnum" interface of libudev has been disabled, as
it was generally incompatible with device namespacing as
sequence numbers of devices go "missing" if the devices are
part of a different namespace.
* "systemctl list-timers" and "systemctl list-sockets" gained
a --recursive switch for showing units of these types also
for all local containers, similar in style to the already
supported --recursive switch for "systemctl list-units".
* A new RebootArgument= setting has been added for service
units, which may be used to specify a kernel reboot argument
to use when triggering reboots with StartLimitAction=.
* A new FailureAction= setting has been added for service
units which may be used to specify an operation to trigger
when a service fails. This works similarly to
StartLimitAction=, but unlike it, controls what is done
immediately rather than only after several attempts to
restart the service in question.
* hostnamed got updated to also expose the kernel name,
release, and version on the bus. This is useful for
executing commands like hostnamectl with the -H switch.
systemd-analyze makes use of this to properly display
details when running non-locally.
* The bootchart tool can now show cgroup information in the
graphs it generates.
* The CFS CPU quota cgroup attribute is now exposed for
services. The new CPUQuota= switch has been added for this
which takes a percentage value. Setting this will have the
result that a service may never get more CPU time than the
specified percentage, even if the machine is otherwise idle.
* systemd-networkd learned IPIP and SIT tunnel support.
* LSB init scripts exposing a dependency on $network will now
get a dependency on rather than simply This should bring LSB handling closer to
what it was on SysV systems.
* A new kernel option has been added to control
how fsck shall deal with unclean file systems at boot.
* The (.ini) configuration file parser will now silently
ignore sections whose name begins with "X-". This may be
used to maintain application-specific extension sections in unit
* machined gained a new API to query the IP addresses of
registered containers. "machinectl status" has been updated
to show these addresses in its output.
* A new call sd_uid_get_display() has been added to the
sd-login APIs for querying the "primary" session of a
user. The "primary" session of the user is elected from the
user's sessions and generally a graphical session is
preferred over a text one.
* A minimal systemd-resolved daemon has been added. It
currently simply acts as a companion to systemd-networkd and
manages resolv.conf based on per-interface DNS
configuration, possibly supplied via DHCP. In the long run
we hope to extend this into a local DNSSEC enabled DNS and
mDNS cache.
* The systemd-networkd-wait-online tool is now enabled by
default. It will delay until a network
connection has been configured. The tool primarily integrates
with networkd, but will also make a best effort to make sense
of network configuration performed in some other way.
* Two new service options StartupCPUShares= and
StartupBlockIOWeight= have been added that work similarly to
CPUShares= and BlockIOWeight= however only apply during
system startup. This is useful to prioritize certain services
differently during bootup than during normal runtime.
* hostnamed has been changed to prefer the statically
configured hostname in /etc/hostname (unless set to
'localhost' or empty) over any dynamic one supplied by
dhcp. With this change, the rules for picking the hostname
match more closely the rules of other configuration settings
where the local administrator's configuration in /etc always
overrides any other settings.
Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
Lindskog, WaLyong Cho, Will Woods, Zbigniew
-- Beijing, 2014-05-28
* When restoring the screen brightness at boot, stay away from
the darkest setting or from the lowest 5% of the available
range, depending on which is the larger value of both. This
should effectively protect the user from rebooting into a
black screen, should the brightness have been set to minimum
by accident.
* sd-login gained a new sd_machine_get_class() call to
determine the class ("vm" or "container") of a machine
registered with machined.
* sd-login gained new calls
to query the identity of the peer of a local AF_UNIX
connection. They operate similarly to their sd_pid_get_xyz()
* PID 1 will now maintain a system-wide system state engine
with the states "starting", "running", "degraded",
"maintenance", "stopping". These states are bound to system
startup, normal runtime, runtime with at least one failed
service, rescue/emergency mode and system shutdown. This
state is shown in the "systemctl status" output when no unit
name is passed. It is useful to determine system state, in
particularly when doing so for many systems or containers at
* A new command "list-machines" has been added to "systemctl"
that lists all local OS containers and shows their system
state (see above), if systemd runs inside of them.
* systemctl gained a new "-r" switch to recursively enumerate
units on all local containers, when used with the
"list-unit" command (which is the default one that is
executed when no parameters are specified).
* The GPT automatic partition discovery logic will now honour
two GPT partition flags: one may be set on a partition to
cause it to be mounted read-only, and the other may be set
on a partition to ignore it during automatic discovery.
* Two new GPT type UUIDs have been added for automatic root
partition discovery, for 32-bit and 64-bit ARM. This is not
particularly useful for discovering the root directory on
these architectures during bare-metal boots (since UEFI is
not common there), but still very useful to allow booting of
ARM disk images in nspawn with the -i option.
* MAC addresses of interfaces created with nspawn's
--network-interface= switch will now be generated from the
machine name, and thus be stable between multiple invocations
of the container.
* logind will now automatically remove all IPC objects owned
by a user if she or he fully logs out. This makes sure that
users who are logged out cannot continue to consume IPC
resources. This covers SysV memory, semaphores and message
queues as well as POSIX shared memory and message
queues. Traditionally, SysV and POSIX IPC had no life-cycle
limits. With this functionality, that is corrected. This may
be turned off by using the RemoveIPC= switch of logind.conf.
* The systemd-machine-id-setup and tmpfiles tools gained a
--root= switch to operate on a specific root directory,
instead of /.
* journald can now forward logged messages to the TTYs of all
logged in users ("wall"). This is the default for all
emergency messages now.
* A new tool systemd-journal-remote has been added to stream
journal log messages across the network.
* /sys/fs/cgroup/ is now mounted read-only after all cgroup
controller trees are mounted into it. Note that the
directories mounted beneath it are not read-only. This is a
security measure and is particularly useful because glibc
actually includes a search logic to pick any tmpfs it can
find to implement shm_open() if /dev/shm is not available
(which it might very well be in namespaced setups).
* machinectl gained a new "poweroff" command to cleanly power
down a local OS container.
* The PrivateDevices= unit file setting will now also drop the
CAP_MKNOD capability from the capability bound set, and
imply DevicePolicy=closed.
* PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
comprehensively on all long-running systemd services where
this is appropriate.
* systemd-udevd will now run in a disassociated mount
namespace. To mount directories from udev rules, make sure to
pull in mount units via SYSTEMD_WANTS properties.
* The kdbus support gained support for uploading policy into
the kernel. sd-bus gained support for creating "monitoring"
connections that can eavesdrop into all bus communication
for debugging purposes.
* Timestamps may now be specified in seconds since the UNIX
epoch Jan 1st, 1970 by specifying "@" followed by the value
in seconds.
* Native tcpwrap support in systemd has been removed. tcpwrap
is old code, not really maintained anymore and has serious
shortcomings, and better options such as firewalls
exist. For setups that require tcpwrap usage, please
consider invoking your socket-activated service via tcpd,
like on traditional inetd.
* A new system.conf configuration option
DefaultTimerAccuracySec= has been added that controls the
default AccuracySec= setting of .timer units.
* Timer units gained a new WakeSystem= switch. If enabled,
timers configured this way will cause the system to resume
from system suspend (if the system supports that, which most
do these days).
* Timer units gained a new Persistent= switch. If enabled,
timers configured this way will save to disk when they have
been last triggered. This information is then used on next
reboot to possible execute overdue timer events, that
could not take place because the system was powered off.
This enables simple anacron-like behaviour for timer units.
* systemctl's "list-timers" will now also list the time a
timer unit was last triggered in addition to the next time
it will be triggered.
* systemd-networkd will now assign predictable IPv4LL
addresses to its local interfaces.
Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
-- Berlin, 2014-03-25
* A new unit file setting RestrictAddressFamilies= has been
added to restrict which socket address families unit
processes gain access to. This takes address family names
like "AF_INET" or "AF_UNIX", and is useful to minimize the
attack surface of services via exotic protocol stacks. This
is built on seccomp system call filters.
* Two new unit file settings RuntimeDirectory= and
RuntimeDirectoryMode= have been added that may be used to
manage a per-daemon runtime directories below /run. This is
an alternative for setting up directory permissions with
tmpfiles snippets, and has the advantage that the runtime
directory's lifetime is bound to the daemon runtime and that
the daemon starts up with an empty directory each time. This
is particularly useful when writing services that drop
privileges using the User= or Group= setting.
* The DeviceAllow= unit setting now supports globbing for
matching against device group names.
* The systemd configuration file system.conf gained new
settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
DefaultMemoryAccounting= to globally turn on/off accounting
for specific resources (cgroups) for all units. These
settings may still be overridden individually in each unit
* systemd-gpt-auto-generator is now able to discover /srv and
root partitions in addition to /home and swap partitions. It
also supports LUKS-encrypted partitions now. With this in
place, automatic discovery of partitions to mount following
the Discoverable Partitions Specification
is now a lot more complete. This allows booting without
/etc/fstab and without root= on the kernel command line on
systems prepared appropriately.
* systemd-nspawn gained a new --image= switch which allows
booting up disk images and Linux installations on any block
device that follow the Discoverable Partitions Specification
(see above). This means that installations made with
appropriately updated installers may now be started and
deployed using container managers, completely
unmodified. (We hope that libvirt-lxc will add support for
this feature soon, too.)
* systemd-nspawn gained a new --network-macvlan= setting to
set up a private macvlan interface for the
container. Similarly, systemd-networkd gained a new
Kind=macvlan setting in .netdev files.
* systemd-networkd now supports configuring local addresses
using IPv4LL.
* A new tool systemd-network-wait-online has been added to
synchronously wait for network connectivity using
* The sd-bus.h bus API gained a new sd_bus_track object for
tracking the life-cycle of bus peers. Note that sd-bus.h is
still not a public API though (unless you specify
--enable-kdbus on the configure command line, which however
voids your warranty and you get no API stability guarantee).
* The $XDG_RUNTIME_DIR runtime directories for each user are
now individual tmpfs instances, which has the benefit of
introducing separate pools for each user, with individual
size limits, and thus making sure that unprivileged clients
can no longer negatively impact the system or other users by
filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
RuntimeDirectorySize= has been introduced that allows
controlling the default size limit for all users. It
defaults to 10% of the available physical memory. This is no
replacement for quotas on tmpfs though (which the kernel
still does not support), as /dev/shm and /tmp are still
shared resources used by both the system and unprivileged
* logind will now automatically turn off automatic suspending
on laptop lid close when more than one display is
connected. This was previously expected to be implemented
individually in desktop environments (such as GNOME),
however has been added to logind now, in order to fix a
boot-time race where a desktop environment might not have
been started yet and thus not been able to take an inhibitor
lock at the time where logind already suspends the system
due to a closed lid.
* logind will now wait at least 30s after each system
suspend/resume cycle, and 3min after system boot before
suspending the system due to a closed laptop lid. This
should give USB docking stations and similar enough time to
be probed and configured after system resume and boot in
order to then act as suspend blocker.
* systemd-run gained a new --property= setting which allows
initialization of resource control properties (and others)
for the created scope or service unit. Example: "systemd-run
--property=BlockIOWeight=10 updatedb" may be used to run
updatedb at a low block IO scheduling weight.
* systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
now also work in --scope mode.
* When systemd is compiled with kdbus support, basic support
for enforced policies is now in place. (Note that enabling
kdbus still voids your warranty and no API compatibility
promises are made.)
Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
Zbigniew Jędrzejewski-Szmek
-- Berlin, 2014-03-12
* systemd will now relabel /dev after loading the SMACK policy
according to SMACK rules.
* A new unit file option AppArmorProfile= has been added to
set the AppArmor profile for the processes of a unit.
* A new condition check ConditionArchitecture= has been added
to conditionalize units based on the system architecture, as
reported by uname()'s "machine" field.
* systemd-networkd now supports matching on the system
virtualization, architecture, kernel command line, host name
and machine ID.
* logind is now a lot more aggressive when suspending the
machine due to a closed laptop lid. Instead of acting only
on the lid close action, it will continuously watch the lid
status and act on it. This is useful for laptops where the
power button is on the outside of the chassis so that it can
be reached without opening the lid (such as the Lenovo
Yoga). On those machines, logind will now immediately
re-suspend the machine if the power button has been
accidentally pressed while the laptop was suspended and in a
backpack or similar.
* logind will now watch SW_DOCK switches and inhibit reaction
to the lid switch if it is pressed. This means that logind
will not suspend the machine anymore if the lid is closed
and the system is docked, if the laptop supports SW_DOCK
notifications via the input layer. Note that ACPI docking
stations do not generate this currently. Also note that this
logic is usually not fully sufficient and Desktop
Environments should take a lid switch inhibitor lock when an
external display is connected, as systemd will not watch
this on its own.
* nspawn will now make use of the devices cgroup controller by
default, and only permit creation of and access to the usual
API device nodes like /dev/null or /dev/random, as well as
access to (but not creation of) the pty devices.
* We will now ship a default .network file for
systemd-networkd that automatically configures DHCP for
network interfaces created by nspawn's --network-veth or
--network-bridge= switches.
* systemd will now understand the usual M, K, G, T suffixes
according to SI conventions (i.e. to the base 1000) when
referring to throughput and hardware metrics. It will stay
with IEC conventions (i.e. to the base 1024) for software
metrics, according to what is customary according to
Wikipedia. We explicitly document which base applies for
each configuration option.
* The DeviceAllow= setting in unit files now supports a syntax
to whitelist an entire group of devices node majors at once,
based on the /proc/devices listing. For example, with the
string "char-pts", it is now possible to whitelist all
current and future pseudo-TTYs at once.
* sd-event learned a new "post" event source. Event sources of
this type are triggered by the dispatching of any event
source of a type that is not "post". This is useful for
implementing clean-up and check event sources that are
triggered by other work being done in the program.
* systemd-networkd is no longer statically enabled, but uses
the usual [Install] sections so that it can be
enabled/disabled using systemctl. It still is enabled by
default however.
* When creating a veth interface pair with systemd-nspawn, the
host side will now be prefixed with "vb-" if
--network-bridge= is used, and with "ve-" if --network-veth
is used. This way, it is easy to distinguish these cases on
the host, for example to apply different configuration to
them with systemd-networkd.
* The compatibility libraries for,, and do not make use of IFUNC
anymore. Instead, we now build multiple times
under these alternative names. This means that the footprint
is drastically increased, but given that these are
transitional compatibility libraries, this should not matter
much. This change has been made necessary to support the ARM
platform for these compatibility libraries, as the ARM
toolchain is not really at the same level as the toolchain
for other architectures like x86 and does not support
IFUNC. Please make sure to use --enable-compat-libs only
during a transitional period!
Contributions from: Andreas Fuchs, Armin K., Colin Walters,
Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
Zbigniew Jędrzejewski-Szmek
-- Berlin, 2014-02-24
* A new component "systemd-networkd" has been added that can
be used to configure local network interfaces statically or
via DHCP. It is capable of bringing up bridges, VLANs, and
bonding. Currently, no hook-ups for interactive network
configuration are provided. Use this for your initrd,
container, embedded, or server setup if you need a simple,
yet powerful, network configuration solution. This
configuration subsystem is quite nifty, as it allows wildcard
hotplug matching in interfaces. For example, with a single
configuration snippet, you can configure that all Ethernet
interfaces showing up are automatically added to a bridge,
or similar. It supports link-sensing and more.
* A new tool "systemd-socket-proxyd" has been added which can
act as a bidirectional proxy for TCP sockets. This is
useful for adding socket activation support to services that
do not actually support socket activation, including virtual
machines and the like.
* Add a new tool to save/restore rfkill state on
* Save/restore state of keyboard backlights in addition to
display backlights on shutdown/boot.
* udev learned a new SECLABEL{} construct to label device
nodes with a specific security label when they appear. For
now, only SECLABEL{selinux} is supported, but the syntax is
prepared for additional security frameworks.
* udev gained a new scheme to configure link-level attributes
from files in /etc/systemd/network/*.link. These files can
match against MAC address, device path, driver name and type,
and will apply attributes like the naming policy, link speed,
MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
address assignment policy (randomized, ...).
* The configuration of network interface naming rules for
"permanent interface names" has changed: a new NamePolicy=
setting in the [Link] section of .link files determines the
priority of possible naming schemes (onboard, slot, mac,
path). The default value of this setting is determined by
/usr/lib/net/links/ Old
80-net-name-slot.rules udev configuration file has been
removed, so local configuration overriding this file should
be adapated to override instead.
* When the User= switch is used in a unit file, also
initialize $SHELL= based on the user database entry.
* systemd no longer depends on libdbus. All communication is
now done with sd-bus, systemd's low-level bus library
* kdbus support has been added to PID 1 itself. When kdbus is
enabled, this causes PID 1 to set up the system bus and
enable support for a new ".busname" unit type that
encapsulates bus name activation on kdbus. It works a little
bit like ".socket" units, except for bus names. A new
generator has been added that converts classic dbus1 service
activation files automatically into native systemd .busname
and .service units.
* sd-bus: add a light-weight vtable implementation that allows
defining objects on the bus with a simple static const
vtable array of its methods, signals and properties.
* systemd will not generate or install static dbus
introspection data anymore to /usr/share/dbus-1/interfaces,
as the precise format of these files is unclear, and
nothing makes use of it.
* A proxy daemon is now provided to proxy clients connecting
via classic D-Bus AF_UNIX sockets to kdbus, to provide full
compatibility with classic D-Bus.
* A bus driver implementation has been added that supports the
classic D-Bus bus driver calls on kdbus, also for
compatibility purposes.
* A new API "sd-event.h" has been added that implements a
minimal event loop API built around epoll. It provides a
couple of features that direct epoll usage is lacking:
prioritization of events, scales to large numbers of timer
events, per-event timer slack (accuracy), system-wide
coalescing of timer events, exit handlers, watchdog
supervision support using systemd's sd_notify() API, child
process handling.
* A new API "sd-rntl.h" has been added that provides an API
around the route netlink interface of the kernel, similar in
style to "sd-bus.h".
* A new API "sd-dhcp-client.h" has been added that provides a
small DHCPv4 client-side implementation. This is used by
* There is a new kernel command line option
"systemd.restore_state=0|1". When set to "0", none of the
systemd tools will restore saved runtime state to hardware
devices. More specifically, the rfkill and backlight states
are not restored.
* The FsckPassNo= compatibility option in mount/service units
has been removed. The fstab generator will now add the
necessary dependencies automatically, and does not require
PID1's support for that anymore.
* journalctl gained a new switch, --list-boots, that lists
recent boots with their times and boot IDs.
* The various tools like systemctl, loginctl, timedatectl,
busctl, systemd-run, ... have gained a new switch "-M" to
connect to a specific, local OS container (as direct
connection, without requiring SSH). This works on any
container that is registered with machined, such as those
created by libvirt-lxc or nspawn.
* systemd-run and systemd-analyze also gained support for "-H"
to connect to remote hosts via SSH. This is particularly
useful for systemd-run because it enables queuing of jobs
onto remote systems.
* machinectl gained a new command "login" to open a getty
login in any local container. This works with any container
that is registered with machined (such as those created by
libvirt-lxc or nspawn), and which runs systemd inside.
* machinectl gained a new "reboot" command that may be used to
trigger a reboot on a specific container that is registered
with machined. This works on any container that runs an init
system of some kind.
* systemctl gained a new "list-timers" command to print a nice
listing of installed timer units with the times they elapse
* Alternative reboot() parameters may now be specified on the
"systemctl reboot" command line and are passed to the
reboot() system call.
* systemctl gained a new --job-mode= switch to configure the
mode to queue a job with. This is a more generic version of
--fail, --irreversible, and --ignore-dependencies, which are
still available but not advertised anymore.
* /etc/systemd/system.conf gained new settings to configure
various default timeouts of units, as well as the default
start limit interval and burst. These may still be overridden
within each Unit.
* PID1 will now export on the bus profile data of the security
policy upload process (such as the SELinux policy upload to
the kernel).
* journald: when forwarding logs to the console, include
timestamps (following the setting in
* OnCalendar= in timer units now understands the special
strings "yearly" and "annually". (Both are equivalent)
* The accuracy of timer units is now configurable with the new
AccuracySec= setting. It defaults to 1min.
* A new dependency type JoinsNamespaceOf= has been added that
allows running two services within the same /tmp and network
namespace, if PrivateNetwork= or PrivateTmp= are used.
* A new command "cat" has been added to systemctl. It outputs
the original unit file of a unit, and concatenates the
contents of additional "drop-in" unit file snippets, so that
the full configuration is shown.
* systemctl now supports globbing on the various "list-xyz"
commands, like "list-units" or "list-sockets", as well as on
those commands which take multiple unit names.
* journalctl's --unit= switch gained support for globbing.
* All systemd daemons now make use of the watchdog logic so
that systemd automatically notices when they hang.
* If the $container_ttys environment variable is set,
getty-generator will automatically spawn a getty for each
listed tty. This is useful for container managers to request
login gettys to be spawned on as many ttys as needed.
* %h, %s, %U specifier support is not available anymore when
used in unit files for PID 1. This is because NSS calls are
not safe from PID 1. They stay available for --user
instances of systemd, and as special case for the root user.
* loginctl gained a new "--no-legend" switch to turn off output
of the legend text.
* The "sd-login.h" API gained three new calls:
sd_session_is_remote(), sd_session_get_remote_user(),
sd_session_get_remote_host() to query information about
remote sessions.
* The udev hardware database now also carries vendor/product
information of SDIO devices.
* The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
determine whether watchdog notifications are requested by
the system manager.
* Socket-activated per-connection services now include a
short description of the connection parameters in the
* tmpfiles gained a new "--boot" option. When this is not used,
only lines where the command character is not suffixed with
"!" are executed. When this option is specified, those
options are executed too. This partitions tmpfiles
directives into those that can be safely executed at any
time, and those which should be run only at boot (for
example, a line that creates /run/nologin).
* A new API "sd-resolve.h" has been added which provides a simple
asynchronous wrapper around glibc NSS host name resolution
calls, such as getaddrinfo(). In contrast to glibc's
getaddrinfo_a(), it does not use signals. In contrast to most
other asynchronous name resolution libraries, this one does
not reimplement DNS, but reuses NSS, so that alternate
host name resolution systems continue to work, such as mDNS,
LDAP, etc. This API is based on libasyncns, but it has been
cleaned up for inclusion in systemd.
* The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
"sd-daemon.h" are no longer found in individual libraries,,, Instead, we have
merged them into a single library,, which
provides all symbols. The reason for this is cyclic
dependencies, as these libraries tend to use each other's
symbols. So far, we have managed to workaround that by linking
a copy of a good part of our code into each of these
libraries again and again, which, however, makes certain
things hard to do, like sharing static variables. Also, it
substantially increases footprint. With this change, there
is only one library for the basic APIs systemd
provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
"sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
library as well, however are subject to the --enable-kdbus
switch (see below). Note that "sd-dhcp-client.h" is not part
of this library (this is because it only consumes, never
provides, services of/to other APIs). To make the transition
easy from the separate libraries to the unified one, we
provide the --enable-compat-libs compile-time switch which
will generate stub libraries that are compatible with the
old ones but redirect all calls to the new one.
* All of the kdbus logic and the new APIs "sd-bus.h",
"sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
and "sd-utf8.h" are compile-time optional via the
"--enable-kdbus" switch, and they are not compiled in by
default. To make use of kdbus, you have to explicitly enable
the switch. Note however, that neither the kernel nor the
userspace API for all of this is considered stable yet. We
want to maintain the freedom to still change the APIs for
now. By specifying this build-time switch, you acknowledge
that you are aware of the instability of the current
* Also, note that while kdbus is pretty much complete,
it lacks one thing: proper policy support. This means you
can build a fully working system with all features; however,
it will be highly insecure. Policy support will be added in
one of the next releases, at the same time that we will
declare the APIs stable.
* When the kernel command-line argument "kdbus" is specified,
systemd will automatically load the kdbus.ko kernel module. At
this stage of development, it is only useful for testing kdbus
and should not be used in production. Note: if "--enable-kdbus"
is specified, and the kdbus.ko kernel module is available, and
"kdbus" is added to the kernel command line, the entire system
runs with kdbus instead of dbus-daemon, with the above mentioned
problem of missing the system policy enforcement. Also a future
version of kdbus.ko or a newer systemd will not be compatible with
each other, and will unlikely be able to boot the machine if only
one of them is updated.
* systemctl gained a new "import-environment" command which
uploads the caller's environment (or parts thereof) into the
service manager so that it is inherited by services started
by the manager. This is useful to upload variables like
$DISPLAY into the user service manager.
* A new PrivateDevices= switch has been added to service units
which allows running a service with a namespaced /dev
directory that does not contain any device nodes for
physical devices. More specifically, it only includes devices
such as /dev/null, /dev/urandom, and /dev/zero which are API
entry points.
* logind has been extended to support behaviour like VT
switching on seats that do not support a VT. This makes
multi-session available on seats that are not the first seat
(seat0), and on systems where kernel support for VTs has
been disabled at compile-time.
* If a process holds a delay lock for system sleep or shutdown
and fails to release it in time, we will now log its
identity. This makes it easier to identify processes that
cause slow suspends or power-offs.
* When parsing /etc/crypttab, support for a new key-slot=
option as supported by Debian is added. It allows indicating
which LUKS slot to use on disk, speeding up key loading.
* The sd_journald_sendv() API call has been checked and
officially declared to be async-signal-safe so that it may
be invoked from signal handlers for logging purposes.
* Boot-time status output is now enabled automatically after a
short timeout if boot does not progress, in order to give
the user an indication what she or he is waiting for.
* The boot-time output has been improved to show how much time
remains until jobs expire.
* The KillMode= switch in service units gained a new possible
value "mixed". If set, and the unit is shut down, then the
initial SIGTERM signal is sent only to the main daemon
process, while the following SIGKILL signal is sent to
all remaining processes of the service.
* When a scope unit is registered, a new property "Controller"
may be set. If set to a valid bus name, systemd will send a
RequestStop() signal to this name when it would like to shut
down the scope. This may be used to hook manager logic into
the shutdown logic of scope units. Also, scope units may now
be put in a special "abandoned" state, in which case the
manager process which created them takes no further
responsibilities for it.
* When reading unit files, systemd will now verify
the access mode of these files, and warn about certain
suspicious combinations. This has been added to make it
easier to track down packaging bugs where unit files are
marked executable or world-writable.
* systemd-nspawn gained a new "--setenv=" switch to set
container-wide environment variables. The similar option in
systemd-activate was renamed from "--environment=" to
"--setenv=" for consistency.
* systemd-nspawn has been updated to create a new kdbus domain
for each container that is invoked, thus allowing each
container to have its own set of system and user buses,
independent of the host.
* systemd-nspawn gained a new --drop-capability= switch to run
the container with less capabilities than the default. Both
--drop-capability= and --capability= now take the special
string "all" for dropping or keeping all capabilities.
* systemd-nspawn gained new switches for executing containers
with specific SELinux labels set.
* systemd-nspawn gained a new --quiet switch to not generate
any additional output but the container's own console
* systemd-nspawn gained a new --share-system switch to run a
container without PID namespacing enabled.
* systemd-nspawn gained a new --register= switch to control
whether the container is registered with systemd-machined or
not. This is useful for containers that do not run full
OS images, but only specific apps.
* systemd-nspawn gained a new --keep-unit which may be used
when invoked as the only program from a service unit, and
results in registration of the unit service itself in
systemd-machined, instead of a newly opened scope unit.
* systemd-nspawn gained a new --network-interface= switch for
moving arbitrary interfaces to the container. The new
--network-veth switch creates a virtual Ethernet connection
between host and container. The new --network-bridge=
switch then allows assigning the host side of this virtual
Ethernet connection to a bridge device.
* systemd-nspawn gained a new --personality= switch for
setting the kernel personality for the container. This is
useful when running a 32-bit container on a 64-bit host. A
similar option Personality= is now also available for service
units to use.
* logind will now also track a "Desktop" identifier for each
session which encodes the desktop environment of it. This is
useful for desktop environments that want to identify
multiple running sessions of itself easily.
* A new SELinuxContext= setting for service units has been
added that allows setting a specific SELinux execution
context for a service.
* Most systemd client tools will now honour $SYSTEMD_LESS for
settings of the "less" pager. By default, these tools will
override $LESS to allow certain operations to work, such as
jump-to-the-end. With $SYSTEMD_LESS, it is possible to
influence this logic.
* systemd's "seccomp" hook-up has been changed to make use of
the libseccomp library instead of using its own
implementation. This has benefits for portability among
other things.
* For usage together with SystemCallFilter=, a new
SystemCallErrorNumber= setting has been introduced that
allows configuration of a system error number to be returned
on filtered system calls, instead of immediately killing the
process. Also, SystemCallArchitectures= has been added to
limit access to system calls of a particular architecture
(in order to turn off support for unused secondary
architectures). There is also a global
SystemCallArchitectures= setting in system.conf now to turn
off support for non-native system calls system-wide.
* systemd requires a kernel with a working name_to_handle_at(),
please see the kernel config requirements in the README file.
Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
Elia Pinto, Florian Weimer, George McCollister, Goffredo
Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
-- Berlin, 2014-02-20
* logind has gained support for facilitating privileged input
and drm device access for unprivileged clients. This work is
useful to allow Wayland display servers (and similar
programs, such as kmscon) to run under the user's ID and
access input and drm devices which are normally
protected. When this is used (and the kernel is new enough)
logind will "mute" IO on the file descriptors passed to
Wayland as long as it is in the background and "unmute" it
if it returns into the foreground. This allows secure
session switching without allowing background sessions to
eavesdrop on input and display data. This also introduces
session switching support if VT support is turned off in the
kernel, and on seats that are not seat0.
* A new kernel command line option luks.options= is understood
now which allows specifiying LUKS options for usage for LUKS
encrypted partitions specified with luks.uuid=.
* tmpfiles.d(5) snippets may now use specifier expansion in
path names. More specifically %m, %b, %H, %v, are now
replaced by the local machine id, boot id, hostname, and
kernel version number.
* A new tmpfiles.d(5) command "m" has been introduced which
may be used to change the owner/group/access mode of a file
or directory if it exists, but do nothing if it does not.
* This release removes high-level support for the
MemorySoftLimit= cgroup setting. The underlying kernel
cgroup attribute memory.soft_limit= is currently badly
designed and likely to be removed from the kernel API in its
current form, hence we should not expose it for now.
* The memory.use_hierarchy cgroup attribute is now enabled for
all cgroups systemd creates in the memory cgroup
hierarchy. This option is likely to be come the built-in
default in the kernel anyway, and the non-hierarchial mode
never made much sense in the intrinsically hierarchial
cgroup system.
* A new field _SYSTEMD_SLICE= is logged along with all journal
messages containing the slice a message was generated
from. This is useful to allow easy per-customer filtering of
logs among other things.
* systemd-journald will no longer adjust the group of journal
files it creates to the "systemd-journal" group. Instead we
rely on the journal directory to be owned by the
"systemd-journal" group, and its setgid bit set, so that the
kernel file system layer will automatically enforce that
journal files inherit this group assignment. The reason for
this change is that we cannot allow NSS look-ups from
journald which would be necessary to resolve
"systemd-journal" to a numeric GID, because this might
create deadlocks if NSS involves synchronous queries to
other daemons (such as nscd, or sssd) which in turn are
logging clients of journald and might block on it, which
would then dead lock. A tmpfiles.d(5) snippet included in
systemd will make sure the setgid bit and group are
properly set on the journal directory if it exists on every
boot. However, we recommend adjusting it manually after
upgrades too (or from RPM scriptlets), so that the change is
not delayed until next reboot.
* Backlight and random seed files in /var/lib/ have moved into
the /var/lib/systemd/ directory, in order to centralize all
systemd generated files in one directory.
* Boot time performance measurements (as displayed by
"systemd-analyze" for example) will now read ACPI 5.0 FPDT
performance information if that's available to determine how
much time BIOS and boot loader initialization required. With
a sufficiently new BIOS you hence no longer need to boot
with Gummiboot to get access to such information.
Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
Cristian Rodríguez, Dave Reisner, David Herrmann, David
Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
-- Berlin, 2013-10-02
* The Restart= option for services now understands a new
on-watchdog setting, which will restart the service
automatically if the service stops sending out watchdog keep
alive messages (as configured with WatchdogSec=).
* The getty generator (which is responsible for bringing up a
getty on configured serial consoles) will no longer only
start a getty on the primary kernel console but on all
others, too. This makes the order in which console= is
specified on the kernel command line less important.
* libsystemd-logind gained a new sd_session_get_vt() call to
retrieve the VT number of a session.
* If the option "tries=0" is set for an entry of /etc/crypttab
its passphrase is queried indefinitely instead of any
maximum number of tries.
* If a service with a configure PID file terminates its PID
file will now be removed automatically if it still exists
afterwards. This should put an end to stale PID files.
* systemd-run will now also take relative binary path names
for execution and no longer insists on absolute paths.
* InaccessibleDirectories= and ReadOnlyDirectories= now take
paths that are optionally prefixed with "-" to indicate that
it should not be considered a failure if they do not exist.
* journalctl -o (and similar commands) now understands a new
output mode "short-precise", it is similar to "short" but
shows timestamps with usec accuracy.
* The option "discard" (as known from Debian) is now
synonymous to "allow-discards" in /etc/crypttab. In fact,
"discard" is preferred now (since it is easier to remember
and type).
* Some licensing clean-ups were made, so that more code is now
LGPL-2.1 licensed than before.
* A minimal tool to save/restore the display backlight
brightness across reboots has been added. It will store the
backlight setting as late as possible at shutdown, and
restore it as early as possible during reboot.
* A logic to automatically discover and enable home and swap
partitions on GPT disks has been added. With this in place
/etc/fstab becomes optional for many setups as systemd can
discover certain partitions located on the root disk
automatically. Home partitions are recognized under their
GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
partitions are recognized under their GPT type ID
* systemd will no longer pass any environment from the kernel
or initrd to system services. If you want to set an
environment for all services, do so via the kernel command
line systemd.setenv= assignment.
* The systemd-sysctl tool no longer natively reads the file
/etc/sysctl.conf. If desired, the file should be symlinked
from /etc/sysctl.d/99-sysctl.conf. Apart from providing
legacy support by a symlink rather than built-in code, it
also makes the otherwise hidden order of application of the
different files visible. (Note that this partly reverts to a
pre-198 application order of sysctl knobs!)
* The "systemctl set-log-level" and "systemctl dump" commands
have been moved to systemd-analyze.
* systemd-run learned the new --remain-after-exit switch,
which causes the scope unit not to be cleaned up
automatically after the process terminated.
* tmpfiles learned a new --exclude-prefix= switch to exclude
certain paths from operation.
* journald will now automatically flush all messages to disk
as soon as a message of the log priorities CRIT, ALERT or
EMERG is received.
Contributions from: Andrew Cook, Brandon Philips, Christian
Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
William Giokas, Zbigniew Jędrzejewski-Szmek
-- Berlin, 2013-09-13
* The documentation has been updated to cover the various new
concepts introduced with 205.
* Unit files now understand the new %v specifier which
resolves to the kernel version string as returned by "uname
* systemctl now supports filtering the unit list output by
load state, active state and sub state, using the new
--state= parameter.
* "systemctl status" will now show the results of the
condition checks (like ConditionPathExists= and similar) of
the last start attempts of the unit. They are also logged to
the journal.
* "journalctl -b" may now be used to look for boot output of a
specific boot. Try "journalctl -b -1" for the previous boot,
but the syntax is substantially more powerful.
* "journalctl --show-cursor" has been added which prints the
cursor string the last shown log line. This may then be used
with the new "journalctl --after-cursor=" switch to continue
browsing logs from that point on.
* "journalctl --force" may now be used to force regeneration
of an FSS key.
* Creation of "dead" device nodes has been moved from udev
into kmod and tmpfiles. Previously, udev would read the kmod
databases to pre-generate dead device nodes based on meta
information contained in kernel modules, so that these would
be auto-loaded on access rather then at boot. As this
does not really have much to do with the exposing actual
kernel devices to userspace this has always been slightly
alien in the udev codebase. Following the new scheme kmod
will now generate a runtime snippet for tmpfiles from the
module meta information and it now is tmpfiles' job to the
create the nodes. This also allows overriding access and
other parameters for the nodes using the usual tmpfiles
facilities. As side effect this allows us to remove the
CAP_SYS_MKNOD capability bit from udevd entirely.
* logind's device ACLs may now be applied to these "dead"
devices nodes too, thus finally allowing managed access to
devices such as /dev/snd/sequencer whithout loading the
backing module right-away.
* A new RPM macro has been added that may be used to apply
tmpfiles configuration during package installation.
* systemd-detect-virt and ConditionVirtualization= now can
detect User-Mode-Linux machines (UML).
* journald will now implicitly log the effective capabilities
set of processes in the message metadata.
* systemd-cryptsetup has gained support for TrueCrypt volumes.
* The initrd interface has been simplified (more specifically,
support for passing performance data via environment
variables and fsck results via files in /run has been
removed). These features were non-essential, and are
nowadays available in a much nicer way by having systemd in
the initrd serialize its state and have the hosts systemd
deserialize it again.
* The udev "keymap" data files and tools to apply keyboard
specific mappings of scan to key codes, and force-release
scan code lists have been entirely replaced by a udev
"keyboard" builtin and a hwdb data file.
* systemd will now honour the kernel's "quiet" command line
argument also during late shutdown, resulting in a
completely silent shutdown when used.
* There's now an option to control the SO_REUSEPORT socket
option in .socket units.
* Instance units will now automatically get a per-template
subslice of system.slice unless something else is explicitly
configured. For example, instances of sshd@.service will now
implicitly be placed in system-sshd.slice rather than
system.slice as before.
* Test coverage support may now be enabled at build time.
Contributions from: Dave Reisner, Frederic Crozat, Harald
Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
Giokas, Zbigniew Jędrzejewski-Szmek
-- Berlin, 2013-07-23
* Two new unit types have been introduced:
Scope units are very similar to service units, however, are
created out of pre-existing processes -- instead of PID 1
forking off the processes. By using scope units it is
possible for system services and applications to group their
own child processes (worker processes) in a powerful way
which then maybe used to organize them, or kill them
together, or apply resource limits on them.
Slice units may be used to partition system resources in an
hierarchial fashion and then assign other units to them. By
default there are now three slices: system.slice (for all
system services), user.slice (for all user sessions),
machine.slice (for VMs and containers).
Slices and scopes have been introduced primarily in
context of the work to move cgroup handling to a
single-writer scheme, where only PID 1
creates/removes/manages cgroups.
* There's a new concept of "transient" units. In contrast to
normal units these units are created via an API at runtime,
not from configuration from disk. More specifically this
means it is now possible to run arbitrary programs as
independent services, with all execution parameters passed
in via bus APIs rather than read from disk. Transient units
make systemd substantially more dynamic then it ever was,
and useful as a general batch manager.
* logind has been updated to make use of scope and slice units
for managing user sessions. As a user logs in he will get
his own private slice unit, to which all sessions are added
as scope units. We also added support for automatically
adding an instance of user@.service for the user into the
slice. Effectively logind will no longer create cgroup
hierarchies on its own now, it will defer entirely to PID 1
for this by means of scope, service and slice units. Since
user sessions this way become entities managed by PID 1
the output of "systemctl" is now a lot more comprehensive.
* A new mini-daemon "systemd-machined" has been added which
may be used by virtualization managers to register local
VMs/containers. nspawn has been updated accordingly, and
libvirt will be updated shortly. machined will collect a bit
of meta information about the VMs/containers, and assign
them their own scope unit (see above). The collected
meta-data is then made available via the "machinectl" tool,
and exposed in "ps" and similar tools. machined/machinectl
is compile-time optional.
* As discussed earlier, the low-level cgroup configuration
options ControlGroup=, ControlGroupModify=,
ControlGroupPersistent=, ControlGroupAttribute= have been
removed. Please use high-level attribute settings instead as
well as slice units.
* A new bus call SetUnitProperties() has been added to alter
various runtime parameters of a unit. This is primarily
useful to alter cgroup parameters dynamically in a nice way,
but will be extended later on to make more properties
modifiable at runtime. systemctl gained a new set-properties
command that wraps this call.
* A new tool "systemd-run" has been added which can be used to
run arbitrary command lines as transient services or scopes,
while configuring a number of settings via the command
line. This tool is currently very basic, however already
very useful. We plan to extend this tool to even allow
queuing of execution jobs with time triggers from the
command line, similar in fashion to "at".
* nspawn will now inform the user explicitly that kernels with
audit enabled break containers, and suggest the user to turn
off audit.
* Support for detecting the IMA and AppArmor security
frameworks with ConditionSecurity= has been added.
* journalctl gained a new "-k" switch for showing only kernel
messages, mimicking dmesg output; in addition to "--user"
and "--system" switches for showing only user's own logs
and system logs.
* systemd-delta can now show information about drop-in
snippets extending unit files.
* libsystemd-bus has been substantially updated but is still
not available as public API.
* systemd will now look for the "debug" argument on the kernel
command line and enable debug logging, similar to what
"systemd.log_level=debug" already did before.
* "systemctl set-default", "systemctl get-default" has been
added to configure the symlink, which
controls what to boot into by default.
* "systemctl set-log-level" has been added as a convenient
way to raise and lower systemd logging threshold.
* "systemd-analyze plot" will now show the time the various
generators needed for execution, as well as information
about the unit file loading.
* libsystemd-journal gained a new sd_journal_open_files() call
for opening specific journal files. journactl also gained a
new switch to expose this new functionality. Previously we
only supported opening all files from a directory, or all
files from the system, as opening individual files only is
racy due to journal file rotation.
* systemd gained the new DefaultEnvironment= setting in
/etc/systemd/system.conf to set environment variables for
all services.
* If a privileged process logs a journal message with the
OBJECT_PID= field set, then journald will automatically
augment this with additional OBJECT_UID=, OBJECT_GID=,
OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
system services want to log events about specific client
processes. journactl/systemctl has been updated to make use
of this information if all log messages regarding a specific
unit is requested.
Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
Reisner, David Coppa, David King, David Strauss, Eelco
Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
Łukasz Stelmach, 장동준
* The Python bindings gained some minimal support for the APIs
exposed by libsystemd-logind.
* ConditionSecurity= gained support for detecting SMACK. Since
this condition already supports SELinux and AppArmor we only
miss IMA for this. Patches welcome!
Contributions from: Karol Lewandowski, Lennart Poettering,
Zbigniew Jędrzejewski-Szmek
* systemd-nspawn will now create /etc/resolv.conf if
necessary, before bind-mounting the host's file onto it.
* systemd-nspawn will now store meta information about a
container on the container's cgroup as extended attribute
fields, including the root directory.
* The cgroup hierarchy has been reworked in many ways. All
objects any of the components systemd creates in the cgroup
tree are now suffixed. More specifically, user sessions are
now placed in cgroups suffixed with ".session", users in
cgroups suffixed with ".user", and nspawn containers in
cgroups suffixed with ".nspawn". Furthermore, all cgroup
names are now escaped in a simple scheme to avoid collision
of userspace object names with kernel filenames. This work
is preparation for making these objects relocatable in the
cgroup tree, in order to allow easy resource partitioning of
these objects without causing naming conflicts.
* systemctl list-dependencies gained the new switches
--plain, --reverse, --after and --before.
* systemd-inhibit now shows the process name of processes that
have taken an inhibitor lock.
* nss-myhostname will now also resolve "localhost"
implicitly. This makes /etc/hosts an optional file and
nicely handles that on IPv6 ::1 maps to both "localhost" and
the local hostname.
* gained a new call
sd_get_machine_names() to enumerate running containers and
VMs (currently only supported by very new libvirt and
nspawn). sd_login_monitor can now be used to watch
VMs/containers coming and going.
* .include is not allowed recursively anymore, and only in
unit files. Usually it is better to use drop-in snippets in
.d/*.conf anyway, as introduced with systemd 198.
* systemd-analyze gained a new "critical-chain" command that
determines the slowest chain of units run during system
boot-up. It is very useful for tracking down where
optimizing boot time is the most beneficial.
* systemd will no longer allow manipulating service paths in
the name=systemd:/system cgroup tree using ControlGroup= in
units. (But is still fine with it in all other dirs.)
* There's a new systemd-nspawn@.service service file that may
be used to easily run nspawn containers as system
services. With the container's root directory in
/var/lib/container/foobar it is now sufficient to run
"systemctl start systemd-nspawn@foobar.service" to boot it.
* systemd-cgls gained a new parameter "--machine" to list only
the processes within a certain container.
* ConditionSecurity= now can check for "apparmor". We still
are lacking checks for SMACK and IMA for this condition
check though. Patches welcome!
* A new configuration file /etc/systemd/sleep.conf has been
added that may be used to configure which kernel operation
systemd is supposed to execute when "suspend", "hibernate"
or "hybrid-sleep" is requested. This makes the new kernel
"freeze" state accessible to the user.
* ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
the passed argument if applicable.
Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
* The output of 'systemctl list-jobs' got some polishing. The
'--type=' argument may now be passed more than once. A new
command 'systemctl list-sockets' has been added which shows
a list of kernel sockets systemd is listening on with the
socket units they belong to, plus the units these socket
units activate.
* The experimental libsystemd-bus library got substantial
updates to work in conjunction with the (also experimental)
kdbus kernel project. It works well enough to exchange
messages with some sophistication. Note that kdbus is not
ready yet, and the library is mostly an elaborate test case
for now, and not installable.
* systemd gained a new unit 'systemd-static-nodes.service'
that generates static device nodes earlier during boot, and
can run in conjunction with udev.
* libsystemd-login gained a new call sd_pid_get_user_unit()
to retrieve the user systemd unit a process is running
in. This is useful for systems where systemd is used as
session manager.
* systemd-nspawn now places all containers in the new /machine
top-level cgroup directory in the name=systemd
hierarchy. libvirt will soon do the same, so that we get a
uniform separation of /system, /user and /machine for system
services, user processes and containers/virtual
machines. This new cgroup hierarchy is also useful to stick
stable names to specific container instances, which can be
recognized later this way (this name may be controlled
via systemd-nspawn's new -M switch). libsystemd-login also
gained a new call sd_pid_get_machine_name() to retrieve the
name of the container/VM a specific process belongs to.
* bootchart can now store its data in the journal.
* libsystemd-journal gained a new call
sd_journal_add_conjunction() for AND expressions to the
matching logic. This can be used to express more complex
logical expressions.
* journactl can now take multiple --unit= and --user-unit=
* The cryptsetup logic now understands the "luks.key=" kernel
command line switch for specifying a file to read the
decryption key from. Also, if a configured key file is not
found the tool will now automatically fall back to prompting
the user.
* Python systemd.journal module was updated to wrap recently
added functions from libsystemd-journal. The interface was
changed to bring the low level interface in s.j._Reader
closer to the C API, and the high level interface in
s.j.Reader was updated to wrap and convert all data about
an entry.
Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
Tom Gundersen, Zbigniew Jędrzejewski-Szmek
* journalctl --update-catalog now understands a new --root=
option to operate on catalogs found in a different root
* During shutdown after systemd has terminated all running
services a final killing loop kills all remaining left-over
processes. We will now print the name of these processes
when we send SIGKILL to them, since this usually indicates a
* If /etc/crypttab refers to password files stored on
configured mount points automatic dependencies will now be
generated to ensure the specific mount is established first
before the key file is attempted to be read.
* 'systemctl status' will now show information about the
network sockets a socket unit is listening on.
* 'systemctl status' will also shown information about any
drop-in configuration file for units. (Drop-In configuration
files in this context are files such as
* systemd-cgtop now optionally shows summed up CPU times of
cgroups. Press '%' while running cgtop to switch between
percentage and absolute mode. This is useful to determine
which cgroups use up the most CPU time over the entire
runtime of the system. systemd-cgtop has also been updated
to be 'pipeable' for processing with further shell tools.
* 'hostnamectl set-hostname' will now allow setting of FQDN
* The formatting and parsing of time span values has been
changed. The parser now understands fractional expressions
such as "5.5h". The formatter will now output fractional
expressions for all time spans under 1min, i.e. "5.123456s"
rather than "5s 123ms 456us". For time spans under 1s
millisecond values are shown, for those under 1ms
microsecond values are shown. This should greatly improve
all time-related output of systemd.
* libsystemd-login and libsystemd-journal gained new
functions for querying the poll() events mask and poll()
timeout value for integration into arbitrary event
* localectl gained the ability to list available X11 keymaps
(models, layouts, variants, options).
* 'systemd-analyze dot' gained the ability to filter for
specific units via shell-style globs, to create smaller,
more useful graphs. I.e. it is now possible to create simple
graphs of all the dependencies between only target units, or
of all units that Avahi has dependencies with.
Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
* The boot-time readahead implementation for rotating media
will now read the read-ahead data in multiple passes which
consist of all read requests made in equidistant time
intervals. This means instead of strictly reading read-ahead
data in its physical order on disk we now try to find a
middle ground between physical and access time order.
* /etc/os-release files gained a new BUILD_ID= field for usage
on operating systems that provide continuous builds of OS
Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
William Douglas, Zbigniew Jędrzejewski-Szmek
* systemd-python gained an API exposing libsystemd-daemon.
* The SMACK setup logic gained support for uploading CIPSO
security policy.
* Behaviour of PrivateTmp=, ReadWriteDirectories=,
ReadOnlyDirectories= and InaccessibleDirectories= has
changed. The private /tmp and /var/tmp directories are now
shared by all processes of a service (which means
ExecStartPre= may now leave data in /tmp that ExecStart= of
the same service can still access). When a service is
stopped its temporary directories are immediately deleted
(normal clean-up with tmpfiles is still done in addition to
this though).
* By default, systemd will now set a couple of sysctl
variables in the kernel: the safe sysrq options are turned
on, IP route verification is turned on, and source routing
disabled. The recently added hardlink and softlink
protection of the kernel is turned on. These settings should
be reasonably safe, and good defaults for all new systems.
* The predictable network naming logic may now be turned off
with a new kernel command line switch: net.ifnames=0.
* A new libsystemd-bus module has been added that implements a
pretty complete D-Bus client library. For details see:
* journald will now explicitly flush the journal files to disk
at the latest 5min after each write. The file will then also
be marked offline until the next write. This should increase
reliability in case of a crash. The synchronization delay
can be configured via SyncIntervalSec= in journald.conf.
* There's a new unit that can be used
to pull in specific services when at least one remote file
system is to be mounted.
* There are new targets and as
canonical targets to pull user timer and path units in
from. This complements with a similar
purpose for socket units.
* libudev gained a new call udev_device_set_attribute_value()
to set sysfs attributes of a device.
* The udev daemon now sets the default number of worker
processes executed in parallel based on the number of available
CPUs instead of the amount of available RAM. This is supposed
to provide a more reliable default and limit a too aggressive
paralellism for setups with 1000s of devices connected.
Contributions from: Auke Kok, Colin Walters, Cristian
Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
Zbigniew Jędrzejewski-Szmek
* Configuration of unit files may now be extended via drop-in
files without having to edit/override the unit files
themselves. More specifically, if the administrator wants to
change one value for a service file foobar.service he can
now do so by dropping in a configuration snippet into
/etc/systemd/system/foobar.service.d/*.conf. The unit logic
will load all these snippets and apply them on top of the
main unit configuration file, possibly extending or
overriding its settings. Using these drop-in snippets is
generally nicer than the two earlier options for changing
unit files locally: copying the files from
/usr/lib/systemd/system/ to /etc/systemd/system/ and editing
them there; or creating a new file in /etc/systemd/system/
that incorporates the original one via ".include". Drop-in
snippets into these .d/ directories can be placed in any
directory systemd looks for units in, and the usual
overriding semantics between /usr/lib, /etc and /run apply
for them too.
* Most unit file settings which take lists of items can now be
reset by assigning the empty string to them. For example,
normally, settings such as Environment=FOO=BAR append a new
environment variable assignment to the environment block,
each time they are used. By assigning Environment= the empty
string the environment block can be reset to empty. This is
particularly useful with the .d/*.conf drop-in snippets
mentioned above, since this adds the ability to reset list
settings from vendor unit files via these drop-ins.
* systemctl gained a new "list-dependencies" command for
listing the dependencies of a unit recursively.
* Inhibitors are now honored and listed by "systemctl
suspend", "systemctl poweroff" (and similar) too, not only
GNOME. These commands will also list active sessions by
other users.
* Resource limits (as exposed by the various control group
controllers) can now be controlled dynamically at runtime
for all units. More specifically, you can now use a command
like "systemctl set-cgroup-attr foobar.service cpu.shares
2000" to alter the CPU shares a specific service gets. These
settings are stored persistently on disk, and thus allow the
administrator to easily adjust the resource usage of
services with a few simple commands. This dynamic resource
management logic is also available to other programs via the
bus. Almost any kernel cgroup attribute and controller is
* systemd-vconsole-setup will now copy all font settings to
all allocated VTs, where it previously applied them only to
the foreground VT.
* libsystemd-login gained the new sd_session_get_tty() API
* This release drops support for a few legacy or
distribution-specific LSB facility names when parsing init
scripts: $x-display-manager, $mail-transfer-agent,
$mail-transport-agent, $mail-transfer-agent, $smtp,
$null. Also, the unit backing
this has been removed. Distributions which want to retain
compatibility with this should carry the burden for
supporting this themselves and patch support for these back
in, if they really need to. Also, the facilities $syslog and
$local_fs are now ignored, since systemd does not support
early-boot LSB init scripts anymore, and these facilities
are implied anyway for normal services. has
also been removed.
* There are new bus calls on PID1's Manager object for
cancelling jobs, and removing snapshot units. Previously,
both calls were only available on the Job and Snapshot
objects themselves.
* systemd-journal-gatewayd gained SSL support.
* The various "environment" files, such as /etc/locale.conf
now support continuation lines with a backslash ("\") as
last character in the line, similarly in style (but different)
to how this is supported in shells.
* For normal user processes the _SYSTEMD_USER_UNIT= field is
now implicitly appended to every log entry logged. systemctl
has been updated to filter by this field when operating on a
user systemd instance.
* nspawn will now implicitly add the CAP_AUDIT_WRITE and
CAP_AUDIT_CONTROL capabilities to the capabilities set for
the container. This makes it easier to boot unmodified
Fedora systems in a container, which however still requires
audit=0 to be passed on the kernel command line. Auditing in
kernel and userspace is unfortunately still too broken in
context of containers, hence we recommend compiling it out
of the kernel or using audit=0. Hopefully this will be fixed
one day for good in the kernel.
* nspawn gained the new --bind= and --bind-ro= parameters to
bind mount specific directories from the host into the
* nspawn will now mount its own devpts file system instance
into the container, in order not to leak pty devices from
the host into the container.
* systemd will now read the firmware boot time performance
information from the EFI variables, if the used boot loader
supports this, and takes it into account for boot performance
analysis via "systemd-analyze". This is currently supported
only in conjunction with Gummiboot, but could be supported
by other boot loaders too. For details see:
* A new generator has been added that automatically mounts the
EFI System Partition (ESP) to /boot, if that directory
exists, is empty, and no other file system has been
configured to be mounted there.
* logind will now send out PrepareForSleep(false) out
unconditionally, after coming back from suspend. This may be
used by applications as asynchronous notification for
system resume events.
* "systemctl unlock-sessions" has been added, that allows
unlocking the screens of all user sessions at once, similar
to how "systemctl lock-sessions" already locked all users
sessions. This is backed by a new D-Bus call UnlockSessions().
* "loginctl seat-status" will now show the master device of a
seat. (i.e. the device of a seat that needs to be around for
the seat to be considered available, usually the graphics
* tmpfiles gained a new "X" line type, that allows
configuration of files and directories (with wildcards) that
shall be excluded from automatic cleanup ("aging").
* udev default rules set the device node permissions now only
at "add" events, and do not change them any longer with a
later "change" event.
* The log messages for lid events and power/sleep keypresses
now carry a message ID.
* We now have a substantially larger unit test suite, but this
continues to be work in progress.
* udevadm hwdb gained a new --root= parameter to change the
root directory to operate relative to.
* logind will now issue a background sync() request to the kernel
early at shutdown, so that dirty buffers are flushed to disk early
instead of at the last moment, in order to optimize shutdown
times a little.
* A new bootctl tool has been added that is an interface for
certain boot loader operations. This is currently a preview
and is likely to be extended into a small mechanism daemon
like timedated, localed, hostnamed, and can be used by
graphical UIs to enumerate available boot options, and
request boot into firmware operations.
* systemd-bootchart has been relicensed to LGPLv2.1+ to match
the rest of the package. It also has been updated to work
correctly in initrds.
* Policykit previously has been runtime optional, and is now
also compile time optional via a configure switch.
* systemd-analyze has been reimplemented in C. Also "systemctl
dot" has moved into systemd-analyze.
* "systemctl status" with no further parameters will now print
the status of all active or failed units.
* Operations such as "systemctl start" can now be executed
with a new mode "--irreversible" which may be used to queue
operations that cannot accidentally be reversed by a later
job queuing. This is by default used to make shutdown
requests more robust.
* The Python API of systemd now gained a new module for
reading journal files.
* A new tool kernel-install has been added that can install
kernel images according to the Boot Loader Specification:
* Boot time console output has been improved to provide
animated boot time output for hanging jobs.
* A new tool systemd-activate has been added which can be used
to test socket activation with, directly from the command
line. This should make it much easier to test and debug
socket activation in daemons.
* journalctl gained a new "--reverse" (or -r) option to show
journal output in reverse order (i.e. newest line first).
* journalctl gained a new "--pager-end" (or -e) option to jump
to immediately jump to the end of the journal in the
pager. This is only supported in conjunction with "less".
* journalctl gained a new "--user-unit=" option, that works
similarly to "--unit=" but filters for user units rather than
system units.
* A number of unit files to ease adoption of systemd in
initrds has been added. This moves some minimal logic from
the various initrd implementations into systemd proper.
* The journal files are now owned by a new group
"systemd-journal", which exists specifically to allow access
to the journal, and nothing else. Previously, we used the
"adm" group for that, which however possibly covers more
than just journal/log file access. This new group is now
already used by systemd-journal-gatewayd to ensure this
daemon gets access to the journal files and as little else
as possible. Note that "make install" will also set FS ACLs
up for /var/log/journal to give "adm" and "wheel" read
access to it, in addition to "systemd-journal" which owns
the journal files. We recommend that packaging scripts also
add read access to "adm" + "wheel" to /var/log/journal, and
all existing/future journal files. To normal users and
administrators little changes, however packagers need to
ensure to create the "systemd-journal" system group at
package installation time.
* The systemd-journal-gatewayd now runs as unprivileged user
systemd-journal-gateway:systemd-journal-gateway. Packaging
scripts need to create these system user/group at
installation time.
* timedated now exposes a new boolean property CanNTP that
indicates whether a local NTP service is available or not.
* systemd-detect-virt will now also detect xen PVs
* The pstore file system is now mounted by default, if it is
* In addition to the SELinux and IMA policies we will now also
load SMACK policies at early boot.
Contributions from: Adel Gadllah, Aleksander Morgado, Auke
Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
Gundersen, Umut Tezduyar, William Giokas, Zbigniew
Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
* Timer units now support calendar time events in addition to
monotonic time events. That means you can now trigger a unit
based on a calendar time specification such as "Thu,Fri
2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
or fifth day of any month of the year 2013, given that it is
a thursday or friday. This brings timer event support
considerably closer to cron's capabilities. For details on
the supported calendar time specification language see
* udev now supports a number of different naming policies for
network interfaces for predictable names, and a combination
of these policies is now the default. Please see this wiki
document for details:
* Auke Kok's bootchart implementation has been added to the
systemd tree. It is an optional component that can graph the
boot in quite some detail. It is one of the best bootchart
implementations around and minimal in its code and
* nss-myhostname has been integrated into the systemd source
tree. nss-myhostname guarantees that the local hostname
always stays resolvable via NSS. It has been a weak
requirement of systemd-hostnamed since a long time, and
since its code is actually trivial we decided to just
include it in systemd's source tree. It can be turned off
with a configure switch.
* The read-ahead logic is now capable of properly detecting
whether a btrfs file system is on SSD or rotating media, in
order to optimize the read-ahead scheme. Previously, it was
only capable of detecting this on traditional file systems
such as ext4.
* In udev, additional device properties are now read from the
IAB in addition to the OUI database. Also, Bluetooth company
identities are attached to the devices as well.
* In service files %U may be used as specifier that is
replaced by the configured user name of the service.
* nspawn may now be invoked without a controlling TTY. This
makes it suitable for invocation as its own service. This
may be used to set up a simple containerized server system
using only core OS tools.
* systemd and nspawn can now accept socket file descriptors
when they are started for socket activation. This enables
implementation of socket activated nspawn
containers. i.e. think about autospawning an entire OS image
when the first SSH or HTTP connection is received. We expect
that similar functionality will also be added to libvirt-lxc
* journalctl will now suppress ANSI color codes when
presenting log data.
* systemctl will no longer show control group information for
a unit if a the control group is empty anyway.
* logind can now automatically suspend/hibernate/shutdown the
system on idle.
* /etc/machine-info and hostnamed now also expose the chassis
type of the system. This can be used to determine whether
the local system is a laptop, desktop, handset or
tablet. This information may either be configured by the
user/vendor or is automatically determined from ACPI and DMI
information if possible.
* A number of PolicyKit actions are now bound together with
"imply" rules. This should simplify creating UIs because
many actions will now authenticate similar ones as well.
* Unit files learnt a new condition ConditionACPower= which
may be used to conditionalize a unit depending on whether an
AC power source is connected or not, of whether the system
is running on battery power.
* systemctl gained a new "is-failed" verb that may be used in
shell scripts and suchlike to check whether a specific unit
is in the "failed" state.
* The EnvironmentFile= setting in unit files now supports file
globbing, and can hence be used to easily read a number of
environment files at once.
* systemd will no longer detect and recognize specific
distributions. All distribution-specific #ifdeffery has been
removed, systemd is now fully generic and
distribution-agnostic. Effectively, not too much is lost as
a lot of the code is still accessible via explicit configure
switches. However, support for some distribution specific
legacy configuration file formats has been dropped. We
recommend distributions to simply adopt the configuration
files everybody else uses now and convert the old
configuration from packaging scripts. Most distributions
already did that. If that's not possible or desirable,
distributions are welcome to forward port the specific
pieces of code locally from the git history.
* When logging a message about a unit systemd will now always
log the unit name in the message meta data.
* localectl will now also discover system locale data that is
not stored in locale archives, but directly unpacked.
* logind will no longer unconditionally use framebuffer
devices as seat masters, i.e. as devices that are required
to be existing before a seat is considered preset. Instead,
it will now look for all devices that are tagged as
"seat-master" in udev. By default framebuffer devices will
be marked as such, but depending on local systems other
devices might be marked as well. This may be used to
integrate graphics cards using closed source drivers (such
as NVidia ones) more nicely into logind. Note however, that
we recommend using the open source NVidia drivers instead,
and no udev rules for the closed-source drivers will be
shipped from us upstream.
Contributions from: Adam Williamson, Alessandro Crismani, Auke
Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
* udev gained support for loading additional device properties
from an indexed database that is keyed by vendor/product IDs
and similar device identifiers. For the beginning this
"hwdb" is populated with data from the well-known PCI and
USB database, but also includes PNP, ACPI and OID data. In
the longer run this indexed database shall grow into
becoming the one central database for non-essential
userspace device metadata. Previously, data from the PCI/USB
database was only attached to select devices, since the
lookup was a relatively expensive operation due to O(n) time
complexity (with n being the number of entries in the
database). Since this is now O(1), we decided to add in this
data for all devices where this is available, by
default. Note that the indexed database needs to be rebuilt
when new data files are installed. To achieve this you need
to update your packaging scripts to invoke "udevadm hwdb
--update" after installation of hwdb data files. For
RPM-based distributions we introduced the new
%udev_hwdb_update macro for this purpose.
* The Journal gained support for the "Message Catalog", an
indexed database to link up additional information with
journal entries. For further details please check:
The indexed message catalog database also needs to be
rebuilt after installation of message catalog files. Use
"journalctl --update-catalog" for this. For RPM-based
distributions we introduced the %journal_catalog_update
macro for this purpose.
* The Python Journal bindings gained support for the standard
Python logging framework.
* The Journal API gained new functions for checking whether
the underlying file system of a journal file is capable of
properly reporting file change notifications, or whether
applications that want to reflect journal changes "live"
need to recheck journal files continuously in appropriate
time intervals.
* It is now possible to set the "age" field for tmpfiles
entries to 0, indicating that files matching this entry
shall always be removed when the directories are cleaned up.
* coredumpctl gained a new "gdb" verb which invokes gdb
right-away on the selected coredump.
* There's now support for "hybrid sleep" on kernels that
support this, in addition to "suspend" and "hibernate". Use
"systemctl hybrid-sleep" to make use of this.
* logind's HandleSuspendKey= setting (and related settings)
now gained support for a new "lock" setting to simply
request the screen lock on all local sessions, instead of
actually executing a suspend or hibernation.
* systemd will now mount the EFI variables file system by
* Socket units now gained support for configuration of the
SMACK security label.
* timedatectl will now output the time of the last and next
daylight saving change.
* We dropped support for various legacy and distro-specific
concepts, such as insserv, early-boot SysV services
(i.e. those for non-standard runlevels such as 'b' or 'S')
or ArchLinux /etc/rc.conf support. We recommend the
distributions who still need support this to either continue
to maintain the necessary patches downstream, or find a
different solution. (Talk to us if you have questions!)
* Various systemd components will now bypass PolicyKit checks
for root and otherwise handle properly if PolicyKit is not
found to be around. This should fix most issues for
PolicyKit-less systems. Quite frankly this should have been
this way since day one. It is absolutely our intention to
make systemd work fine on PolicyKit-less systems, and we
consider it a bug if something does not work as it should if
PolicyKit is not around.
* For embedded systems it is now possible to build udev and
systemd without blkid and/or kmod support.
* "systemctl switch-root" is now capable of switching root
more than once. I.e. in addition to transitions from the
initrd to the host OS it is now possible to transition to
further OS images from the host. This is useful to implement
offline updating tools.
* Various other additions have been made to the RPM macros
shipped with systemd. Use %udev_rules_update() after
installing new udev rules files. %_udevhwdbdir,
%_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
%_sysctldir are now available which resolve to the right
directories for packages to place various data files in.
* journalctl gained the new --full switch (in addition to
--all, to disable ellipsation for long messages.
Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
* journalctl gained new --since= and --until= switches to
filter by time. It also now supports nice filtering for
units via --unit=/-u.
* Type=oneshot services may use ExecReload= and do the
right thing.
* The journal daemon now supports time-based rotation and
vacuuming, in addition to the usual disk-space based
* The journal will now index the available field values for
each field name. This enables clients to show pretty drop
downs of available match values when filtering. The bash
completion of journalctl has been updated
accordingly. journalctl gained a new switch -F to list all
values a certain field takes in the journal database.
* More service events are now written as structured messages
to the journal, and made recognizable via message IDs.
* The timedated, localed and hostnamed mini-services which
previously only provided support for changing time, locale
and hostname settings from graphical DEs such as GNOME now
also have a minimal (but very useful) text-based client
utility each. This is probably the nicest way to changing
these settings from the command line now, especially since
it lists available options and is fully integrated with bash
* There's now a new tool "systemd-coredumpctl" to list and
extract coredumps from the journal.
* We now install a README each in /var/log/ and
/etc/rc.d/init.d explaining where the system logs and init
scripts went. This hopefully should help folks who go to
that dirs and look into the otherwise now empty void and
scratch their heads.
* When user-services are invoked (by systemd --user) the
$MANAGERPID env var is set to the PID of systemd.
* SIGRTMIN+24 when sent to a --user instance will now result
in immediate termination of systemd.
* gatewayd received numerous feature additions such as a
"follow" mode, for live syncing and filtering.
* browse.html now allows filtering and showing detailed
information on specific entries. Keyboard navigation and
mouse screen support has been added.
* gatewayd/journalctl now supports HTML5/JSON
Server-Sent-Events as output.
* The SysV init script compatibility logic will now
heuristically determine whether a script supports the
"reload" verb, and only then make this available as
"systemctl reload".
* "systemctl status --follow" has been removed, use "journalctl
-u" instead.
* journald.conf's RuntimeMinSize=, PersistentMinSize= settings
have been removed since they are hardly useful to be
* And I'd like to take the opportunity to specifically mention
Zbigniew for his great contributions. Zbigniew, you rock!
Contributions from: Andrew Eikum, Christian Hesse, Colin
Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
Jędrzejewski-Szmek, Сковорода Никита Андреевич
* If /etc/vconsole.conf is non-existent or empty we will no
longer load any console font or key map at boot by
default. Instead the kernel defaults will be left
intact. This is definitely the right thing to do, as no
configuration should mean no configuration, and hard-coding
font names that are different on all archs is probably a bad
idea. Also, the kernel default key map and font should be
good enough for most cases anyway, and mostly identical to
the userspace fonts/key maps we previously overloaded them
with. If distributions want to continue to default to a
non-kernel font or key map they should ship a default
/etc/vconsole.conf with the appropriate contents.
Contributions from: Colin Walters, Daniel J Walsh, Dave
Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
* journalctl gained a new --cursor= switch to show entries
starting from the specified location in the journal.
* We now enforce a size limit on journal entry fields exported
with "-o json" in journalctl. Fields larger than 4K will be
assigned null. This can be turned off with --all.
* An (optional) journal gateway daemon is now available as
"systemd-journal-gatewayd.service". This service provides
access to the journal via HTTP and JSON. This functionality
will be used to implement live log synchronization in both
pull and push modes, but has various other users too, such
as easy log access for debugging of embedded devices. Right
now it is already useful to retrieve the journal via HTTP:
# systemctl start systemd-journal-gatewayd.service
# wget http://localhost:19531/entries
This will download the journal contents in a
/var/log/messages compatible format. The same as JSON:
# curl -H"Accept: application/json" http://localhost:19531/entries
This service is also accessible via a web browser where a
single static HTML5 app is served that uses the JSON logic
to enable the user to do some basic browsing of the
journal. This will be extended later on. Here's an example
screenshot of this app in its current state:
Contributions from: Kay Sievers, Lennart Poettering, Robert
Milasan, Tom Gundersen
* The bash completion logic is now available for journalctl
* We do not mount the "cpuset" controller anymore together with
"cpu" and "cpuacct", as "cpuset" groups generally cannot be
started if no parameters are assigned to it. "cpuset" hence
broke code that assumed it it could create "cpu" groups and
just start them.
* journalctl -f will now subscribe to terminal size changes,
and line break accordingly.
Contributions from: Dave Reisner, Kay Sievers, Lennart
Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
* nspawn will now create a symlink /etc/localtime in the
container environment, copying the host's timezone
setting. Previously this has been done via a bind mount, but
since symlinks cannot be bind mounted this has now been
changed to create/update the appropriate symlink.
* journalctl -n's line number argument is now optional, and
will default to 10 if omitted.
* journald will now log the maximum size the journal files may
take up on disk. This is particularly useful if the default
built-in logic of determining this parameter from the file
system size is used. Use "systemctl status
systemd-journald.service" to see this information.
* The multi-seat X wrapper tool has been stripped down. As X
is now capable of enumerating graphics devices via udev in a
seat-aware way the wrapper is not strictly necessary
anymore. A stripped down temporary stop-gap is still shipped
until the upstream display managers have been updated to
fully support the new X logic. Expect this wrapper to be
removed entirely in one of the next releases.
* HandleSleepKey= in logind.conf has been split up into
HandleSuspendKey= and HandleHibernateKey=. The old setting
is not available anymore. X11 and the kernel are
distuingishing between these keys and we should too. This
also means the inhibition lock for these keys has been split
into two.
Contributions from: Dave Airlie, Eelco Dolstra, Lennart
Poettering, Lukas Nykryn, Václav Pavlín
* Whenever a unit changes state we will now log this to the
journal and show along the unit's own log output in
"systemctl status".
* ConditionPathIsMountPoint= can now properly detect bind
mount points too. (Previously, a bind mount of one file
system to another place in the same file system could not be
detected as mount, since they shared struct stat's st_dev
* We will now mount the cgroup controllers cpu, cpuacct,
cpuset and the controllers net_cls, net_prio together by
* nspawn containers will now have a virtualized boot
ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
over with a randomized ID at container initialization). This
has the effect of making "journalctl -b" do the right thing
in a container.
* The JSON output journal serialization has been updated not
to generate "endless" list objects anymore, but rather one
JSON object per line. This is more in line how most JSON
parsers expect JSON objects. The new output mode
"json-pretty" has been added to provide similar output, but
neatly aligned for readability by humans.
* We dropped all explicit sync() invocations in the shutdown
code. The kernel does this implicitly anyway in the kernel
reboot() syscall. halt(8)'s -n option is now a compatibility
* We now support virtualized reboot() in containers, as
supported by newer kernels. We will fall back to exit() if
CAP_SYS_REBOOT is not available to the container. Also,
nspawn makes use of this now and will actually reboot the
container if the containerized OS asks for that.
* journalctl will only show local log output by default
now. Use --merge (-m) to show remote log output, too.
* libsystemd-journal gained the new sd_journal_get_usage()
call to determine the current disk usage of all journal
files. This is exposed in the new "journalctl --disk-usage"
* journald gained a new configuration setting SplitMode= in
journald.conf which may be used to control how user journals
are split off. See journald.conf(5) for details.
* A new condition type ConditionFileNotEmpty= has been added.
* tmpfiles' "w" lines now support file globbing, to write
multiple files at once.
* We added Python bindings for the journal submission
APIs. More Python APIs for a number of selected APIs will
likely follow. Note that we intend to add native bindings
only for the Python language, as we consider it common
enough to deserve bindings shipped within systemd. There are
various projects outside of systemd that provide bindings
for languages such as PHP or Lua.
* Many conditions will now resolve specifiers such as %i. In
addition, PathChanged= and related directives of .path units
now support specifiers as well.
* There's now a new RPM macro definition for the system preset
dir: %_presetdir.
* journald will now warn if it ca not forward a message to the
syslog daemon because its socket is full.
* timedated will no longer write or process /etc/timezone,
except on Debian. As we do not support late mounted /usr
anymore /etc/localtime always being a symlink is now safe,
and hence the information in /etc/timezone is not necessary
* logind will now always reserve one VT for a text getty (VT6
by default). Previously if more than 6 X sessions where
started they took up all the VTs with auto-spawned gettys,
so that no text gettys were available anymore.
* udev will now automatically inform the btrfs kernel logic
about btrfs RAID components showing up. This should make
simple hotplug based btrfs RAID assembly work.
* PID 1 will now increase its RLIMIT_NOFILE to 64K by default
(but not for its children which will stay at the kernel
default). This should allow setups with a lot more listening
* systemd will now always pass the configured timezone to the
kernel at boot. timedated will do the same when the timezone
is changed.
* logind's inhibition logic has been updated. By default,
logind will now handle the lid switch, the power and sleep
keys all the time, even in graphical sessions. If DEs want
to handle these events on their own they should take the new
handle-power-key, handle-sleep-key and handle-lid-switch
inhibitors during their runtime. A simple way to achiveve
that is to invoke the DE wrapped in an invocation of:
systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
* Access to unit operations is now checked via SELinux taking
the unit file label and client process label into account.
* systemd will now notify the administrator in the journal
when he over-mounts a non-empty directory.
* There are new specifiers that are resolved in unit files,
for the host name (%H), the machine ID (%m) and the boot ID
Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
* Support for reading structured kernel messages from
/dev/kmsg has now been added and is enabled by default.
* Support for reading kernel messages from /proc/kmsg has now
been removed. If you want kernel messages in the journal
make sure to run a recent kernel (>= 3.5) that supports
reading structured messages from /dev/kmsg (see
above). /proc/kmsg is now exclusive property of classic
syslog daemons again.
* The libudev API gained the new
udev_device_new_from_device_id() call.
* The logic for file system namespace (ReadOnlyDirectory=,
ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
require pivot_root() anymore. This means fewer temporary
directories are created below /tmp for this feature.
* nspawn containers will now see and receive all submounts
made on the host OS below the root file system of the
* Forward Secure Sealing is now supported for Journal files,
which provide cryptographical sealing of journal files so
that attackers cannot alter log history anymore without this
being detectable. Lennart will soon post a blog story about
this explaining it in more detail.
* There are two new service settings RestartPreventExitStatus=
and SuccessExitStatus= which allow configuration of exit
status (exit code or signal) which will be excepted from the
restart logic, resp. consider successful.
* journalctl gained the new --verify switch that can be used
to check the integrity of the structure of journal files and
(if Forward Secure Sealing is enabled) the contents of
journal files.
* nspawn containers will now be run with /dev/stdin, /dev/fd/
and similar symlinks pre-created. This makes running shells
as container init process a lot more fun.
* The fstab support can now handle PARTUUID= and PARTLABEL=
* A new ConditionHost= condition has been added to match
against the hostname (with globs) and machine ID. This is
useful for clusters where a single OS image is used to
provision a large number of hosts which shall run slightly
different sets of services.
* Services which hit the restart limit will now be placed in a
failure state.
Contributions from: Bertram Poettering, Dave Reisner, Huang
Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
* When running in --user mode systemd will now become a
subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
tree a lot more organized.
* A new PartOf= unit dependency type has been introduced that
may be used to group services in a natural way.
* "systemctl enable" may now be used to enable instances of
* journalctl now prints error log levels in red, and
warning/notice log levels in bright white. It also supports
filtering by log level now.
* cgtop gained a new -n switch (similar to top), to configure
the maximum number of iterations to run for. It also gained
-b, to run in batch mode (accepting no input).
* The suffix ".service" may now be omitted on most systemctl
command lines involving service unit names.
* There's a new bus call in logind to lock all sessions, as
well as a loginctl verb for it "lock-sessions".
* gained a new call sd_journal_perror()
that works similar to libc perror() but logs to the journal
and encodes structured information about the error number.
* /etc/crypttab entries now understand the new keyfile-size=
* shutdown(8) now can send a (configurable) wall message when
a shutdown is cancelled.
* The mount propagation mode for the root file system will now
default to "shared", which is useful to make containers work
nicely out-of-the-box so that they receive new mounts from
the host. This can be undone locally by running "mount
--make-rprivate /" if needed.
* The prefdm.service file has been removed. Distributions
should maintain this unit downstream if they intend to keep
it around. However, we recommend writing normal unit files
for display managers instead.
* Since systemd is a crucial part of the OS we will now
default to a number of compiler switches that improve
security (hardening) such as read-only relocations, stack
protection, and suchlike.
* The TimeoutSec= setting for services is now split into
TimeoutStartSec= and TimeoutStopSec= to allow configuration
of individual time outs for the start and the stop phase of
the service.
Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
Gundersen, Zbigniew Jędrzejewski-Szmek
* The journal and id128 C APIs are now fully documented as man
* Extra safety checks have been added when transitioning from
the initial RAM disk to the main system to avoid accidental
data loss.
* /etc/crypttab entries now understand the new keyfile-offset=
* systemctl -t can now be used to filter by unit load state.
* The journal C API gained the new sd_journal_wait() call to
make writing synchronous journal clients easier.
* journalctl gained the new -D switch to show journals from a
specific directory.
* journalctl now displays a special marker between log
messages of two different boots.
* The journal is now explicitly flushed to /var via a service
systemd-journal-flush.service, rather than implicitly simply
by seeing /var/log/journal to be writable.
* journalctl (and the journal C APIs) can now match for much
more complex expressions, with alternatives and
* When transitioning from the initial RAM disk to the main
system we will now kill all processes in a killing spree to
ensure no processes stay around by accident.
* Three new specifiers may be used in unit files: %u, %h, %s
resolve to the user name, user home directory resp. user
shell. This is useful for running systemd user instances.
* We now automatically rotate journal files if their data
object hash table gets a fill level > 75%. We also size the
hash table based on the configured maximum file size. This
together should lower hash collisions drastically and thus
speed things up a bit.
* journalctl gained the new "--header" switch to introspect
header data of journal files.
* A new setting SystemCallFilters= has been added to services
which may be used to apply blacklists or whitelists to
system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
* nspawn gained a new --link-journal= switch (and quicker: -j)
to link the container journal with the host. This makes it
very easy to centralize log viewing on the host for all
guests while still keeping the journal files separated.
* Many bugfixes and optimizations
Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
* Several tools now understand kernel command line arguments,
which are only read when run in an initial RAM disk. They
usually follow closely their normal counterparts, but are
prefixed with rd.
* There's a new tool to analyze the readahead files that are
automatically generated at boot. Use:
/usr/lib/systemd/systemd-readahead analyze /.readahead
* We now provide an early debug shell on tty9 if this enabled. Use:
systemctl enable debug-shell.service
* All plymouth related units have been moved into the Plymouth
package. Please make sure to upgrade your Plymouth version
as well.
* systemd-tmpfiles now supports getting passed the basename of
a configuration file only, in which case it will look for it
in all appropriate directories automatically.
* udevadm info now takes a /dev or /sys path as argument, and
does the right thing. Example:
udevadm info /dev/sda
udevadm info /sys/class/block/sda
* systemctl now prints a warning if a unit is stopped but a
unit that might trigger it continues to run. Example: a
service is stopped but the socket that activates it is left
* "systemctl status" will now mention if the log output was
shortened due to rotation since a service has been started.
* The journal API now exposes functions to determine the
"cutoff" times due to rotation.
* journald now understands SIGUSR1 and SIGUSR2 for triggering
immediately flushing of runtime logs to /var if possible,
resp. for triggering immediate rotation of the journal
* It is now considered an error if a service is attempted to
be stopped that is not loaded.
* XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
* systemd-analyze now supports Python 3
* tmpfiles now supports cleaning up directories via aging
where the first level dirs are always kept around but
directories beneath it automatically aged. This is enabled
by prefixing the age field with '~'.
* Seat objects now expose CanGraphical, CanTTY properties
which is required to deal with very fast bootups where the
display manager might be running before the graphics drivers
completed initialization.
* Seat objects now expose a State property.
* We now include RPM macros for service enabling/disabling
based on the preset logic. We recommend RPM based
distributions to make use of these macros if possible. This
makes it simpler to reuse RPM spec files across
* We now make sure that the collected systemd unit name is
always valid when services log to the journal via
* There's a new man page kernel-command-line(7) detailing all
command line options we understand.
* The fstab generator may now be disabled at boot by passing
fstab=0 on the kernel command line.
* A new kernel command line option modules-load= is now understood
to load a specific kernel module statically, early at boot.
* Unit names specified on the systemctl command line are now
automatically escaped as needed. Also, if file system or
device paths are specified they are automatically turned
into the appropriate mount or device unit names. Example:
systemctl status /home
systemctl status /dev/sda
* The SysVConsole= configuration option has been removed from
system.conf parsing.
* The SysV search path is no longer exported on the D-Bus
Manager object.
* The Names= option is been removed from unit file parsing.
* There's a new man page bootup(7) detailing the boot process.
* Every unit and every generator we ship with systemd now
comes with full documentation. The self-explanatory boot is
* A couple of services gained "systemd-" prefixes in their
name if they wrap systemd code, rather than only external
code. Among them fsck@.service which is now
* The HaveWatchdog property has been removed from the D-Bus
Manager object.
* systemd.confirm_spawn= on the kernel command line should now
work sensibly.
* There's a new man page crypttab(5) which details all options
we actually understand.
* systemd-nspawn gained a new --capability= switch to pass
additional capabilities to the container.
* timedated will now read known NTP implementation unit names
from /usr/lib/systemd/ntp-units.d/*.list, has been removed.
* journalctl gained a new switch "-b" that lists log data of
the current boot only.
* The notify socket is in the abstract namespace again, in
order to support daemons which chroot() at start-up.
* There is a new Storage= configuration option for journald
which allows configuration of where log data should go. This
also provides a way to disable journal logging entirely, so
that data collected is only forwarded to the console, the
kernel log buffer or another syslog implementation.
* Many bugfixes and optimizations
Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
Shawn Landden, Tom Gundersen
* "systemctl help <unit>" now shows the man page if one is
* Several new man pages have been added.
* MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
MaxLevelConsole= can now be specified in
journald.conf. These options allow reducing the amount of
data stored on disk or forwarded by the log level.
* TimerSlackNSec= can now be specified in system.conf for
PID1. This allows system-wide power savings.
Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
Matthias Clasen
* logind is now capable of (optionally) handling power and
sleep keys as well as the lid switch.
* journalctl now understands the syntax "journalctl
/usr/bin/avahi-daemon" to get all log output of a specific
* CapabilityBoundingSet= in system.conf now also influences
the capability bound set of usermode helpers of the kernel.
Contributions from: Daniel Drake, Daniel J. Walsh, Gert
Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
* Note that we skipped 139 releases here in order to set the
new version to something that is greater than both udev's
and systemd's most recent version number.
* udev: all udev sources are merged into the systemd source tree now.
All future udev development will happen in the systemd tree. It
is still fully supported to use the udev daemon and tools without
systemd running, like in initramfs or other init systems. Building
udev though, will require the *build* of the systemd tree, but
udev can be properly *run* without systemd.
* udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
should be used to create dead device nodes as workarounds for broken
* udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
no longer supported. udev_monitor_new_from_netlink() needs to be
used to subscribe to events.
* udev: when udevd is started by systemd, processes which are left
behind by forking them off of udev rules, are unconditionally cleaned
up and killed now after the event handling has finished. Services or
daemons must be started as systemd services. Services can be
pulled-in by udev to get started, but they can no longer be directly
forked by udev rules.
* udev: the daemon binary is called systemd-udevd now and installed
in /usr/lib/systemd/. Standalone builds or non-systemd systems need
to adapt to that, create symlink, or rename the binary after building
* libudev no longer provides these symbols: