Although JBoss 3.2.x is fairly stable with memory, except for continual application redeployments (I assume this isn't the case for a production system), it could exacerbate any memory leaks. Look through the performance forum for problems that fit your scenario. However, a place to look is at the JDBC drivers (have these changed in the migration?) and the hsqldb (for increasing open network connections). I would also start looking at using a memory profiler to investigate further if nothing becomes apparent.
We are using a new set of JDBC drivers, specifically the latest MSSQL drivers from MS, im hoping to test dropping these back to the older ones to see if this is causing our leak.
The hsqldb is still in use and I don't know why, as we use MSSQL server, why do we still need the hsqldb database setup? I think a different module in jboss uses it and hence has been left in.
Also what memory profilers would you suggest using ? And strategys to use them.. Im looking through the prior posts on memory optimizers and use at the moment.
thanks once again for the help =)
ps if possible these posts should be shifted to the performance forum, my bad for not posting there in the first place.
I think jmemprof is a good open source one. JProbe is good. I like OptimizeIt but also costs. We actually ordered OptimizeIt and ended up using it frequently but jmemprof was handy while we waited for delivery, and turned out to be quite powerful. I'm regretting a bit that we shelled out now since we did solve our problems with jmemprof. YMMV.
I would try to isolate the issue if you can by doing some broad sweeps. The difficulty will lie in being able to reproduce the issue. Some things you may not be able to track so be aware that stored procedures can also have some unexpected effects. I think from past forum issues, there are quite a few that went back to changes in the JDBC drivers.