On Jul 31, 2008, at 4:36 PM, Andrew Thorburn wrote:
Currently playing around with having multiple threads going round and round trying to access the database within Geronimo, essentially to represent multiple users using the application (heavy load), but if I don't have them sleep for long enough, I tend to get errors when they try and access the database (Couldn't get connection within timeout or something), so despite the fact that I've closed the connection it seems as though it's not being released immediately. Or there's possibly something else going on. I also find that my threads just seem to die sometimes, for no obvious reason. They just sit there using up CPU, but not talking to the database or doing any sort of logging or anything. Very strange. I'm probably going to have to add more detail later, but maybe someone has a suggestion?
If there are jta transactions involved then the connections won't be returned to the pool until the tx commits.
Other than that I have no immediate thoughts. Do you have something you could contribute as a sort of stress test?
thanks david jencks
Thanks, - Andrew Thorburn
