I have an apache server that at one point starts to generate many instances to the point that my web page does not load, I have to kill all the apache processes and restart the service, to be able to solve the problems.

This is one of the outputs of the processes

apache 31431 12443 0 09:05? 00:00:12 / usr / sbin / apache2 -k start
apache 31460 12443 0 09:06? 00:00:06 / usr / sbin / apache2 -k start
apache 31502 12443 0 09:10? 00:00:06 / usr / sbin / apache2 -k start
apache 31530 12443 0 09:15? 00:00:08 / usr / sbin / apache2 -k start
apache 31567 12443 0 09:24? 00:00:03 / usr / sbin / apache2 -k start
apache 31576 12443 0 09:24? 00:00:07 / usr / sbin / apache2 -k start
apache 31628 12443 0 09:28? 00:00:03 / usr / sbin / apache2 -k start
apache 31639 12443 0 09:28? 00:00:07 / usr / sbin / apache2 -k start
apache 31642 12443 0 09:28? 00:00:05 / usr / sbin / apache2 -k start
apache 31655 12443 0 09:29? 00:00:05 / usr / sbin / apache2 -k start
apache 31669 12443 0 09:30? 00:00:05 / usr / sbin / apache2 -k start
apache 31707 12443 0 09:38? 00:00:03 / usr / sbin / apache2 -k start
apache 31708 12443 0 09:38? 00:00:02 / usr / sbin / apache2 -k start
apache 31719 12443 0 09:39? 00:00:01 / usr / sbin / apache2 -k start
apache 31728 12443 0 09:40? 00:00:04 / usr / sbin / apache2 -k start
apache 31730 12443 0 09:40? 00:00:02 / usr / sbin / apache2 -k start
apache 31740 12443 0 09:40? 00:00:02 / usr / sbin / apache2 -k start
apache 31747 12443 0 09:40? 00:00:04 / usr / sbin / apache2 -k start
apache 31753 12443 0 09:41? 00:00:01 / usr / sbin / apache2 -k start
apache 31757 12443 0 09:41? 00:00:03 / usr / sbin / apache2 -k start
apache 31775 12443 0 09:42? 00:00:02 / usr / sbin / apache2 -k start

This is the configuration of the workers module
****<IfModule mpm_worker_module>
*******StartServers 128
*******ServerLimit 256
*******MinSpareThreads 32
*******MaxSpareThreads 128
*******ThreadLimit 2048
*******ThreadsPerChild 64
*******MaxClients 4096
*******MaxRequestsPerChild 8000
****</ IfModule>

We do not know what this server is supposed to provide, what the application documentation (if any) suggests performance-wise or what causes the problem without you posting more details.


No it isn't because UNIX doesn't do spaces like that. Minor nit but please post actual representations of what output system commands offer.


What are these settings based on? Application (as in HTTPd) installation defaults? Recommended by the application?
*The first reflex would be to up any values. That's nice but it's wrong. Try to set more aggressive settings: less servers, less spares, shorter TTL's, less requests-per-child. Basically less of everything, forcing the system to work harder. Before you do that establish a baseline by stress testing the application and jotting down configuration changes and the effect they have on performance running consecutive tests.