[Mimedefang] some graphs

[Cor Bosman]

> >First one is Emails/Hour, next is Load, third is CPU.
> >
> >As you can see, around 12:00 at the beginning I took a tiny mailflow and
> >added some to it (by diverting emailflow in a loadbalancer).
> >It's peaking at about 12.000 emails per hour. That's about 3 emails per
> >second. At that point CPU isnt stressed, load is still below 1, hovering
> >around 0.6/0.7. The peaks are burst of traffic I havent really checked.
> >I personally find this an acceptable level with room to spare for
> >sudden abnormalities.
> 
> Why does the load average approach one while the CPU is down in the 20% 
> range? Where's the bottleneck at this load?

No idea actually. Plenty of CPU, no diskio, plenty of memory (1+ GB free).
Nothing obvious to have processes be in a run queue. The only thing
I can imagine is outgoing network traffic. I noticed some xdelays
in outgoing sendmail. Maybe the receiving end is slightly overloaded,

Cor