How To Achieve Accurate Utilisation Readings?

[Ockingshay]

Generally my server runs very well and CPU utilisation is usually well under 10%. I have however noticed that when sickrage is copying a file across after processing, the CPU utilisation jumps up well into the 90s.

 

This obviously causes other things to slow down, so i would like to try and figure out what would be best to improve the situation. I have an intel xeon E3-1220 V2 @ 3.10GHz that is no slouch compared to what we use to run on NAS' and so i wouldn't expect this spike in what is suppose to be a "bread & butter" operation.

 

I have used htop and dynamix webgui to compare processes and they vary wildly. What is the best method for really knowing what my server is doing at a particular time?

 

It happens as well when transmission is running at full speed.

 

I also read that you should have enough RAM to cover the file size but i have 8GB , which again seems a lot for a NAS.

 

I'm also not sure if this is simply normal behaviour and i've just never noticed it before.

 

[Ockingshay]

blimey boys and girls, keep it down in here, i can barely hear myself think!

 

In the interest of sharing knowledge, i negated this problem by using cpu pinning, as (sort of!) described in this thread: http://lime-technology.com/forum/index.php?topic=36257.0

 

By controlling what CPUs my unruly dockers used, i was able to stop them overrunning my system.

 

The guide is a little out of date unless you read the entire thread, so to summarise, you will need to edit each/any docker, click the advanced slider and create an entry in the "extra parameters" field with:

 

--cpuset-cpus=

 

...where after the = is the desired core you would like.

 

you can use ranges i.e. 0-2 or specific cores i.e. 0,1.

 

An example would be to use the 3rd core, noting that core 1 is actually 0 and not 1:

 

--cpuset-cpus=2

 

good luck!

 

PS To answer my original question "How To Achieve Accurate Utilisation Readings?" HTOP showed the true representation.