users@glassfish.java.net

Re: Cluster performance and failover

From: <glassfish_at_javadesktop.org>
Date: Tue, 21 Aug 2007 19:42:40 PDT

I have been looking over this thread. You mentioned in an earlier message that you didn't have problems with simpler apps. Can you try to find out and report just how much data you are replicating in these different scenarios? In general, http session is designed for small amounts of conversational state. It is not good to have large amounts of data. It would be good if you could get a sense of the size of the object graph(s) you are either deliberately (or not deliberately, accidentally??) replicating when using apps built with these UI frameworks you spoke of.

Another point. In an earlier thread you saw a log message saying something took 3+ seconds. That was a load call and essentially it timed out. Is it possible that your load balancer is not supporting cookies correctly. We make use of jsessionid and jsessionid_version cookies.

Also what kind of load are you applying? Is this just a one user functional test? Or something else?
[Message sent by forum member 'lwhite' (lwhite)]

http://forums.java.net/jive/thread.jspa?messageID=231921