Shared publicly  - 
I spent some time yesterday and today to bring the +systemd documentation up-to-date regarding systemd's new cgroup APIs. It's terse but should be quite comprehensive.

The background of this: the kernel cgroup maintainer +Tejun Heo wants userspace to move to a single-writer interface for manipulating the cgroup tree. This means that userspace needs some daemon that takes exclusive control of the cgroup tree, and that all other userspace software needs to talk to it to make use of cgroups, instead of individually interfering directly with cgroupfs. On systemd systems that daemon is going to systemd's PID 1 itself, accessible via D-Bus calls.

This new document reiterates the reasons for all of this, and the gives a terse introduction on the new concepts for resource control in systemd and how to make use of them.

(And in case you wonder: if you don't use systemd, this is irrelevant for you. This is about systemd's APIs for resource control, and that's it. While your non-systemd OS needs to come up with some solution for the single-writer requirement of the  reworked kernel cgroup logic, that solution is not coming from us, and your OS needs to do its homework on its own.)
Starting with version 205 systemd provides a number of interfaces that may be used to create and manage labelled groups of processes for the purpose of monitoring and controlling them and their resource usage. This is built on top of the Linux kernel Control Groups ("cgroups") facility.
Joshua Hillerup's profile photoKyungmin Park's profile photoDiego Call.'s profile photoElia Pinto's profile photo
Add a comment...