Okay
Well, Knight can handle Tons of work.
It was doing alot. BEing a reverse proxy for 3+ Million requests a week. But not for the Z vhost.
It was hitting the [link|http://66.80.246.91/apache2-default/|66.80.246.91 default].
It was logging and getting the proxy requests. 3+ Million a week.
A few days ago Apache had nearly all the memory on knight tied up, I think the 14th. I've also been seeing some screwy things earlier I couldn't explain. I had a few messages in the system logs complaining about some dropped connections.
Many different things.
So in effect Apache was tanking up on the Memory. The work knight was doing was memory intensive, not very CPU intensive... swap was creeping up. SLOWING DOWN any requests or work needed to be down.
The drop from blocking the other networks in December... they were using z as the proxy host vs the open one on the default vhost. So it made a HUGE effect on Zs performance... but I now see it was in conjunction with the other stuff.
We were doing huge numbers, for proxy requests.
