[Devel] [patch rh7 2/2] cgroup: Mangle cgroups root from inside of VE view

[Vladimir Davydov]

On Fri, May 29, 2015 at 11:30:00AM +0300, Cyrill Gorcunov wrote:
> On Fri, May 29, 2015 at 11:18:52AM +0300, Vladimir Davydov wrote:
> > Hi Cyrill,
> > 
> > On Tue, May 26, 2015 at 06:00:52PM +0300, Cyrill Gorcunov wrote:
> > > We're bindmounting cgroups for container so if say a container
> > > is having CTID=200 then @cgroups and @mountinfo output will
> > > contain /200 as a root. Which makes Docker to lookup for
> > > appropriate directory inside /sys/fs/cgroup/<controller>
> > > which of course not present because of been bindmounted
> > > from the node (note we can't bindmount into
> > > <controller>/<container> here because it confuses container's
> > > systemd instance and it stuck on boot).
> > 
> > How is it supposed to be sorted out upstream? Does it mean that systemd
> > simply cannot run inside a container?
> 
> No it can and it does run inside container (I notice some problems
> though, regardless of how we represent cgroup paths inside container).
> The main reason is to make /proc/pid/cgroup output to match what is
> container sees from inside. Or you mean something else?

I mean vanilla kernels, where there is no /proc/pid/cgroup mangling. How
does it work there (if it does)? May be, we could adapt the technique
used there instead of patching the kernel.