I'm not a Java programmer or SQL guy, so forgive my dimness on the subjects. I am network support guy looking at this from the hardware/network side .
I have a Windows 2000 server hosting a Jboss portal (jboss-portal). It accesses a database on a Microsoft SQL Cluster.
Now and then, the developers complain about a strange database connectivity issue between Jboss and the SQL cluster. The issue being is the connections between the two do not close. The claim is that while they are monitoring the connections open and close between the two, things are fine 90% of the time. But, a point is often reached where the connections remain open and then accumulate, passing the maximum allowed connections, requiring the Jboss server to have its jboss-portal service restarted. This is happening almost daily.
From browsing the JBoss directory, it looks like ?jboss-tomcat-4.0.2?. From a external scan, it shows ?Apache Tomcat/Coyote JSP engine 1.1?
Hardware wise, the database is on a SAN. Dozens of other applications have there databases hosted on a SAN without incident, so not sure where to look there unless JBoss has some sensitivities. Another application makes use of the same database without incident (though it is a IIS/ASP app)
What could be the cause of this or where could I have a DBA or JBoss developer investigate if it?s an admin/code issue? (e.g. perfmon has lots of DB counters, but I?m not sure what options I would set it too or perhaps better DB drivers on the app. dev. side)
Failure to severe a networked DB connection untill it crashes sounds more like a memory leak for sockets than anything.
Would anyone concure that this is a code issue rather than a JBoss service or Database issue?