I've a website written in CFML, hosted on a Amazon ec2 istance with Apache and Railo... I'm trying to improve the performace and reduce the request calls.
Previously I had some problems with CFML query of querys (qoq) that generated eccessive reading operation on the disk... But I've solved removing the use of qoq.
Reading the Apache's and FusionReactor's logs i see an high and instantly CPU peak in relation with the request calls... This peak is repeated every hour also in not used moments of the website, like overnight.
I see that request on apache logs and don't understand why are generated:
66.85.140.110 - - [30/Jun/2014:04:10:50 +0200] "GET / HTTP/1.0" 200 -
66.85.144.237 - - [30/Jun/2014:04:50:06 +0200] "GET / HTTP/1.0" 200 -
Soo... I think that is a server configuration that do something every hour, also because in a parallel website on a different host I haven't this peak. I've find this on the apache documentation: https://wiki.apache.org/httpd/InternalDummyConnection
What I'm asking is if this CPU peak can be produced by an Apache internal request or something other server configuration like an wrong JVM parameters setting. An other idea is this peak can be generate by the JVM Garbace Collector when start to clean the memory... There's a way to specify the refresh time of GC?
Someone have a some suggestions?
The two IP's you posted are from securedservers.com - an external source and if I had to guess, it's probably some sort of uptime monitor. You could always block those IP's from your firewall if they're causing you grief and costing you extra money.
On a side note, if Amazon is costing you so much money that it's worth your time to track down tiny little issues like this in order to reduce your costs, then maybe it's time to look at a dedicated server somewhere where you don't have to worry about getting charged for CPU time. Just a thought.