Hello all

I'm running a load test in a portal application that is using jetspeed framewrok. The load test tool used is Load Runner from
mercury interactive. I also used OptimizeIt during the load test sometimes to identify some hot spot areas.
We are trying to run 50concurrent users in a powerful solaris box with 2 CPU's, heap size set to 500MB and the sleep time between the concurrent user requests are 30seconds and the users stay in the system doing activities for 10mts during the load.


I had two questions based on the symptoms I saw during the load test.


1) In JetspeedResources.properties.


#The following property is used for DatabasePsmlManagerService? only. The default is false that is caching is
#not allowed with DatabasePsmlManagerService?, as the current implementation of caching does not work on a
#distributed environment. It works only in a single server environment.
services.PsmlManager?.caching-on=true
services.PsmlManager.cacheSize=50


When I turn on the above property, the performance increased drastically and the memory increased as well. But at some point when I hit the 20 users concurrent, it breaks the entire unix process that is running the application server and the whole application dies. Not sure this is a bug and anyone encountered this. If the value is set to false , then the test goes through fine, but the round trip performance is very slow.
Appreciate if there are any information on what is the implementation behind this.


2) what is the value to be set in the below properties for the above load test scenario.

In JetspeedResources?.properties
Everything under the ThreadPool? Service heading -- Setting these up to the default seemed to adversely affect performance.
services.ThreadPool?.init.count=5
services.ThreadPool?.max.count=20
services.ThreadPool?.minspare.count=5



Appreciate any information on the above questions and thanks a lot in advance.


- Shan

Reply via email to