On Fri, Apr 05, 2013 at 10:16:43AM +0800, Gao feng wrote:
On 2013/03/27 13:26, Gao feng wrote:
> On 2013/03/20 16:14, Gao feng wrote:
>> There are 3 reason we need to rework the cgroupfs
>> mounting in container.
>>
>> 1, Yin Olivia reported a "failed to mount cgroup"
>> problem, now we given that the name of cgroup mount point
>> is same with the subsystem type, Or libvirt_lxc
>> will fail to start.
>>
>> 2, The cgroup configuration is leaked to the container,
>> even user can change host's cgroup configuration in
>> container.
>>
>> 3, After we enable userns, the cgroupfs is unable to be
>> mounted in uninit-userns.
>>
>> This patch tries to resolve these 3 problem,
>> uses mount --bind to set cgroupfs for container.
>>
>> It means the directory /sys/fs/cgroup/memory/libvirt/lxc/domain
>> of host will be binded to the directory /sys/fs/cgroup/memory of
>> container.
>>
>
> what's your idea about this patch?
>
Ping Again
The pach has the right idea, but it clashes with the refactoring I've
done for cgroups and LXC. I'll update your patch to apply ontop of this
series:
https://www.redhat.com/archives/libvir-list/2013-April/msg00352.html
and copy you on the mail when i post it.
Regards,
Daniel
--
|:
http://berrange.com -o-
http://www.flickr.com/photos/dberrange/ :|
|:
http://libvirt.org -o-
http://virt-manager.org :|
|:
http://autobuild.org -o-
http://search.cpan.org/~danberr/ :|
|:
http://entangle-photo.org -o-
http://live.gnome.org/gtk-vnc :|