[Techtalk] Logging server load

[Rasjid Wilcox]

I have recently got myself a virtual box (user-mode-linux) at Linode.com.

It comes with root access, and a very nice front end that means it is almost 
as good as having physical access to the server.  (ie, I can reboot my 
virtual box into a rescue distribution and fix anything I've broken.)  It 
also comes with amazing bandwidth.  I've just donwloaded the java sdk from 
sun at 5.67MB/sec.  (This is bytes, not bits.)  And it is only US$20 per 
month....

Anyway, enough of that.  As you can tell I'm pretty happy with the setup.

Most of the time the response of the server is great.  However, occasionally 
the load goes through the roof.  Well, I exagerate a little - but it gets up 
to 3 or 4.  This is not cpu load (which is still below 10% at the time), just 
the load average.

My guess is that it is something like the locate database update.  My 
experience with user-mode-linux so far is that disk access has quite a high 
overhead.  But I want to generate some logs so I can check this.

Can anyone suggest a better way of logging the load average so I can try and 
work out any pattens, other than:

$ while ((1)); do uptime >> uptime.log ; sleep 60; done

Is there a 'standard' way of doing this?

Cheers,

Rasjid.

-- 
Rasjid Wilcox
Canberra, Australia (UTC +10 hrs)
http://www.openminddev.net