[lxc-devel] A question about lxc

[Daniel Lezcano]

Gui Jianfeng wrote:
> Daniel Lezcano wrote:
>   
>> Gui Jianfeng wrote:
>>     
>>> Hi Daniel,
>>>
>>> I tried lxc and have A question about lxc
>>>
>>> I successfully built lxc in my box, and i played with
>>> it as following.
>>>
>>> #lxc-create -n foo
>>> #lxc-execute -n foo /bin/bash
>>> #lxc-ps -n foo
>>>
>>> Then, I got the following error:
>>>
>>> cat: /usr/local/var/lib/lxc/foo/nsgroup/tasks: No such file or directory
>>> ERROR: List of process IDs must follow -p.
>>>
>>> I checked the code, IMHO, "/usr/local/var/lib/lxc/foo/nsgroup" should
>>> link
>>> to the corresponding Cgroup directory. But i can't find out which part
>>> of the code takes care of creating Cgroup directory.
>>>
>>> Would you explain this a bit?
>>>   
>>>       
>> Hi Gui,
>>
>> lxc does not create the cgroup directory, it is automatically created
>> when a namespace is created, this is a feature of the ns_cgroup. It is
>> probable you mounted the cgroup without the "ns" option or the feature
>> is not enabled in the kernel.
>>
>> There is a tool to check the kernel configuration : lxc-checkconfig; it
>> should give you some informations about the kernel config.
>> You should see something like that:
>>
>> --- Namespaces ---
>> Namespaces: enabled
>> Utsname namespace: enabled
>> Ipc namespace: enabled
>> Pid namespace: enabled
>> User namespace: enabled
>> Network namespace: enabled
>> Multiple /dev/pts instances: enabled
>>
>> --- Control groups ---
>> Cgroup: enabled
>> Cgroup namespace: enabled
>> Cgroup device: enabled
>> Cgroup sched: enabled
>> Cgroup cpu account: enabled
>> Cgroup memory controller: enabled
>> Cgroup cpuset: disabled
>>
>> --- Misc ---
>> Veth pair device: enabled
>> Macvlan: enabled
>>     
>
> Hi Daniel,
>
> I tried lxc-checkconfig and got the following output with the config 
> on 2.6.30.
> It seems "ns_cgroup" is already enabled.
>
> Kernel config /proc/config.gz not found, looking in other places...
> Found kernel config file /lib/modules/2.6.30/build/.config
> --- Namespaces ---
> Namespaces: enabled
> Utsname namespace: enabled
> Ipc namespace: enabled
> Pid namespace: enabled
> User namespace: enabled
> Network namespace: enabled
> Multiple /dev/pts instances: disabled
>
> --- Control groups ---
> Cgroup: enabled
> Cgroup namespace: enabled
> Cgroup device: enabled
> Cgroup sched: disabled
> Cgroup cpu account: enabled
> Cgroup memory controller: enabled
> Cgroup cpuset: enabled
>
> --- Misc ---
> Veth pair device: disabled
> Macvlan: disabled
>   

Can you check in /proc/mount, you have the cgroup mounted with the ns option

Mine is:

/cgroup /cgroup cgroup 
rw,relatime,net_cls,freezer,devices,memory,cpuacct,cpu,ns,debug 0 0

I mounted the cgroup with the option:

mount -t cgroup cgroup /cgroup