305N/Asystemd System and Service Manager
305N/A * udev: all udev sources are merged into the systemd source tree now.
305N/A All future udev development will happen in the systemd tree. It
305N/A is still fully supported to use the udev daemon and tools without
305N/A systemd running, like in initramfs or other init systems. Building
305N/A udev though, will require the *build* of the systemd tree, but
305N/A udev can be properly *run* without systems.
305N/A should be used to create dead device nodes as workarounds for broken
305N/A * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
305N/A no longer supported. udev_monitor_new_from_netlink() needs to be
305N/A used to subscribe to events.
305N/A * udev: when udevd is started by systemd, processes which are left
305N/A behind by forking them off of udev rules, are unconditionally cleaned
305N/A up and killed now after the event handling has finished. Services or
305N/A daemons must be started as systemd services. Services can be
3996N/A pulled-in by udev to ge started, but they can no longer be directly
305N/A * udev: the daemon binary is called systemd-udevd now and installed
305N/A to adapt to that, create symlink, or rename the binary after building
586N/A * systemd-logingctl and systemd-journalctl have been renamed
618N/A to logingctl and journalctl to match systemctl.
305N/A the files to the new names on upgrade.
305N/A * This is mostly a bugfix release
1924N/A * Support optional initialization of the machine ID from the
305N/A KVM or container configured UUID.
844N/A * Support immediate reboots with "systemctl reboot -ff"
3661N/A * Many bugfixes for the journal, including endianess fixes and
3661N/A ensuring that disk space enforcement works
305N/A * We now refuse non-UTF8 strings used in various configuration
305N/A and unit files. This is done to ensure we don't pass invalid
305N/A data over D-Bus or expose it elsewhere.
305N/A * Register Mimo USB Screens as suitable for automatic seat
455N/A * Read SELinux client context from journal clients in a race
305N/A * Reorder configuration file lookup order. /etc now always
305N/A overrides /run in order to allow the administrator to always
305N/A and unconditionally override vendor supplied or
305N/A automatically generated data.
305N/A * The various user visible bits of the journal now have man
305N/A pages. We still lack man pages for the journal API calls
305N/A * We now ship all man pages in HTML format again in the
437N/A Contributions from: Dave Reisner, Dirk Eibach, Frederic
305N/A Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
305N/A Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
305N/A * This is mostly a bugfix release
305N/A * Track class of PAM logins to distuingish greeters from
305N/A Contributions from: Kay Sievers, Lennart Poettering, Michael
1574N/A * This is an important bugfix release for v41.
1574N/A * Building man pages is now optional which should be useful
1574N/A for those building systemd from git but unwilling to install
305N/A * Watchdog support for supervising services is now usable. In
305N/A a future release support for hardware watchdogs
305N/A * Service start rate limiting is now configurable and can be
305N/A turned off per service. When a start rate limit is hit a
3996N/A reboot can automatically be triggered.
3996N/A * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3996N/A Contributions from: Benjamin Franzke, Bill Nottingham,
3996N/A Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3996N/A Schmidt, Michał Górny, Piotr Drąg
An existing
/sbin/init symlink needs to be adapted with the
* The code that loads kernel modules has been ported to invoke
libkmod directly, instead of modprobe. This means we do not
support systems with module-init-tools anymore.
* Watchdog support is now already useful, but still not
understood to set system wide environment variables
* We now limit the set of capabilities of systemd-journald.
* We now set SIGPIPE to ignore by default, since it only is
useful in shell pipelines, and has little use in general
code. This can be disabled with IgnoreSIPIPE=no in unit
Contributions from: Benjamin Franzke, Kay Sievers, Lennart
Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
* This is mostly a bugfix release
* We now expose the reason why a service failed in the
* Rudimentary service watchdog support (will be completed over
* When systemd forks off in order execute some service we will
now immediately changes its argv[0] to reflect which process
it will execute. This is useful to minimize the time window
with a generic argv[0], which makes bootcharts more useful
Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
Mike Kazantsev, Ray Strode
* This is mostly a test release, but incorporates many
* New systemd-cgtop tool to show control groups by their
* Linking against libacl for ACLs is optional again. If
disabled, support tracking device access for active logins
goes becomes unavailable, and so does access to the user
journals by the respective users.
* If a group "adm" exists, journal files are automatically
owned by them, thus allow members of this group full access
to the system journal as well as all user journals.
* The journal now stores the SELinux context of the logging
* Add C++ inclusion guards to all public headers
* New output mode "cat" in the journal to print only text
messages, without any meta data like date or time.
* Include tiny X server wrapper as a temporary stop-gap to
teach XOrg udev display enumeration. This is used by display
managers such as gdm, and will go away as soon as XOrg
learned native udev hotplugging for display devices.
* Add new systemd-cat tool for executing arbitrary programs
BSD logger replacement, and does so by default.
* Optionally store all locally generated coredumps in the
journal along with meta data.
* systemd-tmpfiles learnt four new commands: n, L, c, b, for
writing short strings to files (for usage for /sys), and for
creating symlinks, character and block device nodes.
* New unit file option ControlGroupPersistent= to make cgroups
persistent, following the mechanisms outlined in
* Support multiple local RTCs in a sane way
* No longer monopolize IO when replaying readahead data on
rotating disks, since we might starve non-file-system IO to
death, since fanotify() will not see accesses done by blkid,
* Don't show kernel threads in systemd-cgls anymore, unless
requested with new -k switch.
Contributions from: Dan Horák, Kay Sievers, Lennart
Poettering, Michal Schmidt
* This is mostly a test release, but incorporates many
* The git repository moved to:
* First release with the journal
* The journal replaces both systemd-kmsg-syslogd and
* New sd_pid_get_unit() API call in libsystemd-logind
* Many systemadm clean-ups
remote mounts and may be used to start services before all
* Add bash completion for systemd-loginctl
* Actively monitor PID file creation for daemons which exit in
the parent process before having finished writing the PID
file in the daemon process. Daemons which do this need to be
fixed (
i.e. PID file creation must have finished before the
parent exits), but we now react a bit more gracefully to them.
* Add colourful boot output, mimicking the well-known output
of existing distributions.
* New option PassCredentials= for socket units, for
compatibility with a recent kernel ABI breakage.
thus will no longer act as synchronization point during
* systemctl list-unit-files now supports --root=.
* systemd-tmpfiles now understands two new commands: z, Z for
relabelling files according to the SELinux database. This is
useful to apply SELinux labels to specific files in /sys,
* Output of SysV services is now forwarded to both the console
and the journal by default, not only just the console.
* New man pages for all APIs from libsystemd-login.
* The build tree got reorganized and a the build system is a
lot more modular allowing embedded setups to specifically
select the components of systemd they are interested in.
* Support for Linux systems lacking the kernel VT subsystem is
* configure's --with-rootdir= got renamed to
--with-rootprefix= to follow the naming used by udev and
* Unless specified otherwise we'll now install to /usr instead
* Processes with '@' in argv[0][0] are now excluded from the
final shut-down killing spree, following the logic explained
* All processes remaining in a service cgroup when we enter
the START or START_PRE states are now killed with
SIGKILL. That means it is no longer possible to spawn
background processes from ExecStart= lines (which was never
supported anyway, and bad style).
* New PropagateReloadTo=/PropagateReloadFrom= options to bind
reloading of units together.
Contributions from: Bill Nottingham, Daniel Walsh, Dave
Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek