I made a stress test on the glassfish cluster with apache plugin , just run 300 threads to press , the result is bad ,the average response time is : 20,000ms and the error rate is 30% !
I think is the apache plugin is overload, because , during the test , I visit the service from browse, the response from apache is very slow , and the response from instance directly is fast.
dose anyone get others stress test result ?
is there any way to tune the performance ?
my cluster set :
run hadb for session failover
no.1 machine , 4cpu(Xeon 2G) ,2G mem , apache and instance1 on it
no.2 machine , 4cpu(Xeon 2G),4G mem , instance2 on it
my apache prefork set :
<IfModule prefork.c>
StartServers 1
MinSpareServers 5
MaxSpareServers 10
MaxClients 1
MaxRequestsPerChild 0
</IfModule>
[Message sent by forum member 'dean_chen']
http://forums.java.net/jive/thread.jspa?messageID=393403