On 10/03/2011 11:22 PM, Vivek Goyal wrote:
> On Mon, Oct 03, 2011 at 10:38:20AM +0200, Jan Safranek wrote:
>> On 09/30/2011 05:22 PM, Vivek Goyal wrote:
>>> On Fri, Sep 23, 2011 at 03:52:31PM +0200, Jan Safranek wrote:
>>>> Hello,
>>>>
>>>> here is first bunch of patches that are needed for better cooperation with
>>>> systemd. As agreed, service cgconfig stop/restart should only remove groups
>>>> that cgconfig start has created and only if they are empty. This patchset
>>>> focuses on that - it allows cgclear to parse a config file and remove only
>>>> groups defined there and (optionally) only if they are empty.
>>>>
>>>
>>> Thanks for doing this work Jan.
>>>
>>>> cgclear and cgconfigparser were updated to parse multiple files and even
>>>> directories to simplify the init script and future unit file(s). 
>>>
>>> What's the use case of being able to specify mulitple files?
>>>
> 
> Jan, any thougts on above? Why to providef flexibility to be able to 
> specify a directory of cgconfig files.

The idea is that there is /etc/cgconfig.d/ (or something like that) and
admins or even packages can put their pre-configured groups there. Files
there can have different permissions, so non-root users can still
fine-tune their groups (and only their groups).

> 
>>>>
>>>> The best way to review the set is probably start with the last patch (=man
>>>> pages) so you see the functionality the patchset tries to implement. There 
>>>> are
>>>> also few tests you can examine.
>>>
>>> I see that you introduced an option to cgclear to pass a file. So an init
>>> script is supposed to take snapshot of /etc/cgconfig.conf when cgroup
>>> service starts and then pass that snapshot file to cgclear when cgroup
>>> service is being stopped?
>>
>> Yes, that's the idea for now. Comments are welcome, that's why the
>> series is RFC.
> 
> I think we anyway will do a great deal of things behind "cgconfig"
> service, so I am fine with it. So in fedora 16, cgconfig is still 
> old style service or has been moved to new systemd style service.
> 
>> I'll try to write an unit file and check how it works
>> with systemd and e.g. libvirt.
> 
> That would be good.
> 
>> And I am also working on the sticky bit,
>> but I probably need to break API for that :(.
> 
> Why do we need to break API for that? Due to additional options being
> passed to functions or something else.

Yes, the function which parses a file and creates groups looks only at
the file, now I need to tell it some default permissions. And the
function is in public API. Not a big deal, still I do not like it.

Jan

------------------------------------------------------------------------------
All the data continuously generated in your IT infrastructure contains a
definitive record of customers, application performance, security
threats, fraudulent activity and more. Splunk takes this data and makes
sense of it. Business sense. IT sense. Common sense.
http://p.sf.net/sfu/splunk-d2dcopy1
_______________________________________________
Libcg-devel mailing list
Libcg-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/libcg-devel

Reply via email to