Heya! Let this one be known as the "dynamic" release, where things became dynamic! Or call it the "cgroups" release, where we took possession of the cgroup tree!
This release introduces a number of major new concepts, such as transient units, scopes and slices, which turn systemd into something that is far more dynamic than it ever was (this is primarily made visible in the new "systemd-run" tool, which I invite you to play around with). With this release the systemd binary now does *all* cgroup management (be it as the host's PID1, a session manager, or the PID 1 of a container), and logind and nspawn simply defer their cgroup work. All objects showing up in the cgroup tree are now objects managed by systemd itself. The APIs for this are not documented yet, but will be soon. This brings our systemd userspace much closer to the unified single-writer cgroup hierarchy that Tejun has being working towards from the kernel side. Given that most of the documentation for this is still missing I expect another release soon. Also, there are some other white spots still. Given the lack of documentation this is probably not the release you want to ship your distro with. http://www.freedesktop.org/software/systemd/systemd-205.tar.xz CHANGES WITH 205: * 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 "systemd.log_level=debug" already did before. * "systemctl set-default", "systemctl get-default" has been added to configure the default.target 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 Lennart -- Lennart Poettering - Red Hat, Inc. _______________________________________________ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel
