Okay, I figured out something of what had happened. It *was* a connection
pool problem, and in retrospect, it makes perfect sense: boneheaded move
on my part, partially exposed by someone who rather generously pointed out
some information based on my review of Glassfish (which needs to be
revised, since one of my complaints was based on my own error.)
I made the mistake of thinking the connection pool was a JDBC connection
pool. Bzzt. It was the CONNECTION FACTORY for JMS, being fired off by a
filter which could generate a HUGE number of connections. That's
definitely something in my code, both as something to be fixed in
deployment *and* as something to change about how messages are fired off.
(I need to check the filter to make sure it's not creating a connection
per item, which would be an absurd concept, but hey... new codebase,
entirely untuned.)
Thank you for all who're paying attention. :)
-----------------------------------------------------------------------
Joseph B. Ottinger
http://enigmastation.com
Editor,
http://www.TheServerSide.com joeo@enigmastation.com