I have a server running SUSE. dual quad core chips, 16 gigs of RAM. It's a webserver only running Apache. We have been having a problem the past few days where page loads take forever yet CPU load and RAM load are very, very low. I've noticed when it happens when viewing HTOP that TASKS jump from about 350 average to 1000 average. Pages won't start loading until this number drops. I see no errors in error_log or messages however restarting Apache seems to fix the issue? I've tried messing around with KEEPALIVE settings with no help. Any clues on where to start? Pretty lost on this one becuase I see no errors and the load both CPU and RAM are low.
When watching HTOP another thing I see is all the Apache processes just disappear then after 5 seconds all reappear. Like Apache is dropping all the connections for some reason.
A couple of stats I'd be looking at while this problem is occurring are the load average (not the CPU Load but the load average. Usually appears after the uptime command amongst others) On a quad core box, this shouldn't go above 4 under normal operations. I'd also like to know what your MaxClients setting is in httpd.conf. Lastly, I'd look at the output of "netstat -an" specifically for large numbers of connections in the TIME_WAIT or FIN_WAIT states. These aren't necessarily the problem but hopefully one of them can tell us what the problem is.
When I run netstat -an pretty much all of them are TIME_WAIT, a few are FIN_WAIT1 and a few are ESTABLISHED. The list is huge. What does that mean?