user4318's questions

I have three or four Linux servers, each of which hosts anywhere from 5 to 50 domains. Each domain has its own folder:

/www/projectname/web/

Logs go in:

/www/projectname/log

However, if there's a traffic spike (or, as I see it on my end, a memory usage spike), I'm not sure how to figure out which domain is responsible for the traffic without running tail -f on each of the projects and making an educated guess based on how fast things scroll.

There's got to be a better way! There probably is, but I haven't seen it. And the last time I checked, bandwidth monitors only report system-wide load.

So if anyone knows how to do this the right way, please let me know.

Thanks!

EDIT: My goal is for something that gives instant feedback. I know I can configure Apache to log bytes sent and received, but it doesn't necessarily help if my server is getting hammered and I'm trying to figure out which domain all the traffic is from.

I noticed that one of our Windows XP 32-bit boxes, which has 4GB of RAM, will only use up to 1.3GB of physical memory. Has anyone run into this before?

The only thing I can think of is that it's a 4GB-on-32-bit-OS thing, but I don't see why that would cause XP to only use 1/3 of the RAM.

Thanks!

In answer to your questions:

• The OS displayed 3.25GB of RAM.
• Even under high load (i.e., Photoshop + Firefox + other RAM-intensive programs) the system (according to procexp) only used up to 1.3GB of RAM. RAM load stayed at about 33%.
• I just swapped out one of the 2GB sticks for 1GB, bringing me to 3GB of RAM. Windows XP says I have 3GB of RAM installed. But procexp reports only 750MB being used, or about 25% of RAM.

So it doesn't seem to be a 32/64 bit issue.

The box has a Wolfdale E8400 running on an Asus P5K SE mobo. The OS is Win XP SP3 32-bit.