[jboss-user] [Remoting] - Re: Socket Read Timeout on stressing calls for validating pr

ron.sigal@jboss.com do-not-reply at jboss.com
Wed Jan 14 01:01:30 EST 2009


Hi Vinay,

"vink" wrote : 
  | Issue 1:
  | On providing invokerDestructionDelay=10000, my system run out of threads. I mean my client side. Every call results in the creation of new thread, it seems old invoker thread is not getting reused. 
  | 

Well, I don't think this is a Remoting problem, because the socket transport doesn't create new threads on the client side.  My conjecture is that there is some other problem, and setting "invokerDestructionDelay=10000" is just reducing the consumption of Remoting resources enough to allow this other problem to show up.  What kind of threads are you seeing on the client side?

"vink" wrote : 
  | The best use case for SLSB is to open a socket for a request & close it after finish.
  | 

Actually, that is the default behavior, but it's expensive to recreate sockets.  The "invokerDestructionDelay" parameter is meant to prevent the closing of existing sockets when they are likely to be reused in the near future.  By the way, if a socket is in the TIME_WAIT state, then it *has* been closed.

"vink" wrote : 
  | n the meanwhile, I tried overnight tests on my system with invokerDestructionDelay=10000, & I face this exception on server. I cannot understand what it means, but it is linked with socket; so I thought it may be worth to share;
  | 
  | Caused by: javax.persistence.PersistenceException: org.hibernate.exception.JDBCConnectionException: could not execute query 
  | 

This problem is somewhere between hibernate and your database.  Definitely not a Remoting issue.

-Ron

View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4201580#4201580

Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4201580



More information about the jboss-user mailing list