Reputation: 21
There is an Apache 2 server choking under multiple requests. Amount of processes reaches number set by Maxclient, and new connections pile up in queue. I need to identify which requests take too much time x amount of requests. Apache logs may show either:
%...D The time taken to serve the request, in microseconds.
or
%...T The time taken to serve the request, in seconds.
And this time taken to serve the requests make it's hard to analyze. Because at the time when process table overflows is the time around which I need to measure. And there serve time equals time spent in connection queue plus time processed.
So a .php script which took 2 seconds will show 2 seconds. And .cgi script which spent minute (real data ;)) in queue will show 60.5 seconds.
Is there a way to log time taken to process requests, preferably not involving engaging with hooks?
Upvotes: 2
Views: 6184
Reputation: 2367
Probable this is what you were looking for: https://code.google.com/p/mod-log-firstbyte/
Upvotes: 1