[Users] Getting CPU idle time of CT outside the CT

[Pavel Odintsov]

It's really nice idea for building CPU load graphics from HWN side.
Anybody have any ideas do it without vzctl exec... ?

On Mon, Jul 21, 2014 at 10:53 PM, Todd Mueller <toddmueller at gmail.com> wrote:
> Can you clarify what you are trying to accomplish? Are you trying to
> identify containers using excessive CPU?
>
> On Mon, Jul 21, 2014 at 10:06 AM, Pavel Snajdr <lists at snajpa.net> wrote:
>> Hello,
>>
>> is there any way how to get CPU idle % of CT outside out of the CT?
>>
>> We were trying to figure this out from /proc/vz/vestat, first try was to
>> divide used/idle, but idle seems to be accounted for all available
>> hwnode CPUs and used seems to be accounted for CPUs available to the CT;
>> in the end it was giving a pretty reasonable numbers in low CPU load
>> situations, but when the CT was maxing out its available CPUs, we were
>> getting 100+% CPU load.
>>
>> Next thing we've tried is to look into user,nice and system values, but
>> there is no idle nor total number of available jiffies to compare to.
>>
>> Another thing we've tried is to read /proc/stat from outside of the CT,
>> but that is a problem, since ve->veid is 0 when running a process to
>> read out /vz/root/<veid>/proc/stat.
>>
>> Solution of a last resort seems to be to run mpstat (or equivalent)
>> inside of the CT via vzctl exec and watch its output. But that gives me
>> hundreds of unnecessary running processes - I need to monitor hundreds
>> of CTs per hwnode. That's kind of a pain :)
>>
>> Is there any better way to do this or what am I doing wrong?
>>
>> Our kernel is 042stab092.2.
>>
>> - snajpa
>> _______________________________________________
>> Users mailing list
>> Users at openvz.org
>> https://lists.openvz.org/mailman/listinfo/users
> _______________________________________________
> Users mailing list
> Users at openvz.org
> https://lists.openvz.org/mailman/listinfo/users



-- 
Sincerely yours, Pavel Odintsov