http://httpd.apache.org/docs/2.0/mod/mod_log_config.htmlsays that one of the things you can log in a log file is "The time taken to serve the request, in seconds", however, this counts wall clock time, not CPU time. (I confirmed this using a test script that did nothing but sleep for 20 seconds and exit -- the log file recorded it as taking 20 seconds. to serve.)
I have tried installing mod_status (with ExtendedStatus On) and looking at server_status, but it doesn't appear to display the amount of CPU time that a request has used.
http://httpd.apache.org/docs/2.1/mod/mod_status.htmlsays that one of the "details given" is "The current percentage CPU used by each worker and in total by Apache". However, it seems like this is incorrect -- the "CPU" column doesn't give the percentage of CPU used by the worker, it gives the total CPU time used. (And it's not clear if that refers to CPU time used by that *child*, or by that "slot", or what. I know that column isn't displaying total CPU time used by the current request, because it often displays a number of seconds which is greater than the number of seconds that have passed since a given request was started.)
The only way I can find to identify cpu-intensive requests, is to write a script that grabs the server-status page, then does 'ps auwx' to cross-reference the processes on the server-status page with their pids in 'ps auwx' and get the percentage of CPU usage. (I need both because server-status doesn't show CPU usage, and ps auwx doesn't show the first line of the http request, it just shows the process as /usr/sbin/httpd.) Is there a more straightforward way to do it?
-Bennett --------------------------------------------------------------------- The official User-To-User support forum of the Apache HTTP Server Project. See <URL:http://httpd.apache.org/userslist.html> for more info. To unsubscribe, e-mail: users-unsubscribe@xxxxxxxxxxxxxxxx " from the digest: users-digest-unsubscribe@xxxxxxxxxxxxxxxx For additional commands, e-mail: users-help@xxxxxxxxxxxxxxxx
|