How do I configure LXC to allow the use of SCHED_RR in a container?

Henk asked:

Any ideas on how I can configure LXC/cgroups, to allow the use of SCHED_RR for applications in a container?

On the internet I have seen mention of unmounting the cpu-cgroup, I have tried that, manually (running umount on the cpu cgroup) that did not seem to help.

Any ideas of how I persistently unmount a cgroup? On Fedora 20.

Is unmounting, the cpu cgroup, the correct method for allowing container applications to use the SCHED_RR?

Thanks

My answer:


Remember that typically LXC containers drop capabilities on startup. You need to, at least, allow CAP_SYS_NICE in the container. This capability allows for calling sched_setscheduler() which is the call you need to set SCHED_RR.

An example for reference LXC configuration:

lxc.cap.keep = sys_nice

And of course your applications must also have CAP_SYS_NICE (or be run as root, in which case they already will).


View the full question and any other answers on Server Fault.

Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License.