[systemd-devel] [PATCH] man: document the slice unit

Lennart Poettering lennart at poettering.net
Wed Jul 10 18:02:44 PDT 2013

On Wed, 10.07.13 02:23, Zbigniew Jędrzejewski-Szmek (zbyszek at in.waw.pl) wrote:

> ---
> Does this description fit the masterplan? What about units that
> systemd will expose for CPU and BlockIO shares, will they still be
> 1024 and 1000 based?

Yes, they will.

We probably should split the cgroup stuff out into a separate man page
"systemd.cgroup" or so, similar to the way we already have
"systemd.exec" and "systemd.kill". Note that the cgroup properties are
not only available for slice units but also for service and scope units
(and mount, swap, socket, i.e. everything that forks of processes).

service, socket, swap, mount units all need to reference systemd.exec,
systemd.kill, systemd.cgroup.

scope units need to reference only systemd.kill and
systemd.cgroup. (they do not have any of the settings listed in
systemd.exec, since the processes they contain are created by somebody
else, and hence the settings cannot be applied anymore.)

slice units need to reference only systemd.cgroup. (they contain no
processes directly).

I figure both systemd.slice and systemd.scope will have very short man


Lennart Poettering - Red Hat, Inc.

More information about the systemd-devel mailing list