Best practices for Apache memory and processes under high load
Check whether Apache is running in prefork working mode
apachectl -V | grep MPM
httpd -V | grep MPM
Server MPM: Prefork
-D APACHE_MPM_DIR="server/mpm/prefork"
View the memory currently occupied by each process
top -bn 1 | grep httpd
56961 apache 20 0 406m 49m 40m S 1.9 0.1 0:21.92 httpd
56963 apache 20 0 406m 53m 44m S 1.9 0.1 0:20.45 httpd
56967 apache 20 0 402m 52m 42m S 1.9 0.1 0:20.28 httpd
56971 apache 20 0 404m 54m 44m S 1.9 0.1 0:21.18 httpd
56974 apache 20 0 415m 55m 42m S 1.9 0.1 0:20.07 httpd
RES column is the amount of memory used by the Apache process, the sixth column from the left.
Calculate the number of processes
Maximum number of processes = total amount of memory that Apache can consume/amount of memory consumed by each process
Set related parameters
<IfModule prefork.c>
StartServers 30%MaxClients
MinSpareServers 5%
MaxSpareServers 10%
ServerLimit 256
MaxClients 200 = 10G/50M
MaxRequestsPerChild 10000
</IfModule>
restart
- Apache will actually control the process within this framework. For example, when the load is not high, the process will be automatically reduced.
- The key to apache planning is the amount of memory at maximum load, which is different from nginx
- Calculate the
MaxClientsvalue through the maximum amount of memory
- All other values are calculated based on
MaxClients
