More information about the Underscore mailing list

[_] Tips on System/Performance monitoring for Linode Host

Greg Ferro myetherealmind at gmail.com
Mon Apr 11 12:56:47 BST 2011

Thanks for the tips. htop is several wafts of fulsome joy.

I found a dodgy plugin for a LinkedIn "like" button was the problem and 
now my shiny new Linode ( and free from the turds at MediaTemple ) 
server located in London looks my more content (who cares that 60% of my 
audience is in America)

Hoping for more ad revenue on a faster server.


   1  [||                                                  1.1%]     
Tasks: 53 total, 1 running
   2  [|                                                   0.3%]     
Load average: 0.11 0.44 1.11
   3  [||                                                  1.7%]     
Uptime: 03:54:02
   4  [                                                    0.0%]
   Mem[|||||||||||||||||||||||||||                    254/751MB]
   Swp[|||||                                           19/255MB]
   Time: 12:55:24



-- 
Regards

Greg Ferro
e: myetherealmind at gmail.com
web: http://etherealmind.com | http://packetpushers.net




> ------------------------------------------------------------------------
>
> Amias Channer <mailto:me at amias.net>
> 11 April 2011 12:38
>
>
>
> its worth remembering that linode instances are virtual machines
> so its not so much a problem to max them out than it would be on
> dedicated hardware also load averages are a tricky metric at the
> best of times.
>
> From the top output i'd say its busy but not problematic , the io
> is doing fine as the wait state percentage is 0% , by the looks of its
> just computing something quite heavy or something is not well optimised
> in your setup. This could be your caching plugin generating caches ,
> i've not heard of that one , might be worth trying some others instead.
>
> I'd keep an eye on the load over the next day or so and see if it
> goes down once all the caching kicks in and image manipulation calms
> down. Might be worth trying a stress testing tool on it to see how
> the load increases with lots of requests as this could just be the
> base load and may not increase much more. also more linodes are not
> expensive .
>
> Nagios is an excellent tool for remote monitoring but may be a little
> too complex for your purposes although its definitely worth getting to
> grips with as its pretty much the industry standard and it makes nice
> graphs that help keep managers at bay.
>
> if you need more specific help i'm available on an hourly rate :)
>
> Toodle-pip
> Amias
>
>
>
>
> ------------------------------------------------------------------------
>
> Greg Ferro <mailto:myetherealmind at gmail.com>
> 11 April 2011 11:30
>
>
> I'm running Ubuntu server at Linode and just migrated a couple of my  
> Wordpress blogs there with my main site moving this morning.
>
> However, the CPU utilisation is much higher than I was expecting - 
> typically 50% or more shows in top with a load average above one 
> constantly.
>
> I'm not particularly competent at this level of Linux 'magic hand 
> waving' ( although I migrated my wordpress installs myself so I have 
> some knowledge) and looking for some suggestions for tool / packages  
> that I can use to look into what part of PHP / Apache / MySQL is 
> burning those CPU cycles.
>
> Thanks.
>