I'm running Tomcat 5.5.12 on Mac OS X and have recently switched from
using a Hibernate-configured DBCP to using a JNDI-configured/owned
c3p0 pool. I recognize that this may not necessarily be the correct
place to ask this question, since the problem may be with c3p0, but
perhaps other users may recognize these problems because c3p0 is
commonly used with Tomcat.
The problem I have is that when I redeploy my application, which
itself contains the c3p0 pool <Resource/> in a META-INF/context.xml
file, I eventually reach a point -- after 2 or 3 deploys -- where my
database (PostgreSQL, as it happens) complains about too many clients
connecting.
I am confident that Hibernate is not involved because the problems
begin on redeployment, when the pool tries to acquire and validate
JDBC connections, before Hibernate is given a connection.
Could this perhaps be a failure of Tomcat to un-initialize the pool
on undeployment (which I perform in my build file using Tomcat's Ant
tasks, prior to deployment of the new WAR file)? Or perhaps it is a
failure of c3p0 to release resources when Tomcat asks it to?
I liked that my deployment cycle didn't have to involve restarting
Tomcat, and I think I can probably solve this by putting the c3p0
configuration in <GlobalNamingResources/> in server.xml, but I really
like that my WAR file is totally self-contained right now. It deploys
fine the first time, but subsequent deployments (second or third
time) produce the problem I have outlined here.
Can anyone offer suggestions, or share experiences that could
illuminate a path forward?
Thanks very much.
David
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
