I have a sneaking suspicion that someone is still blowing smoke. ;)  
Either
1) the oracle pool has a leak
2) oracle server has a problem closing connections
3) you have a leak in the application.
 
For problem 3), I find the DBCP's ability to 'tattle' on bad JSP
pages/classes invaluable in tracking down this type of behaviour.  
 
Here's a (big) snip. I've removed a bunch of parameters, as they would
change for your app. But the key ones are included at the bottom.

<Context path="" 
                 docBase="/home/webhome/buzz/" 
                 defaultSessionTimeout="60" 
                 reloadable="true">


<Resource       name="jdbc/BuzzDB"
                auth="Container"
                type="javax.sql.DataSource"     />

<ResourceParams name="jdbc/BuzzDB">

        <parameter>
                <name>factory</name>
        
<value>org.apache.commons.dbcp.BasicDataSourceFactory</value>
        </parameter>

        <!-- Max number of dB connections. Set to 0 for no limit. -->

        <!-- Max number of idle dB connections to retain. Set to 0 for
no limit. -->

                <!-- Max wait for dB connection to become available
(in ms), -1 to wait indefinitely. -->

        <!-- MySQL dB username, password, driver, URL  -->


        <parameter>
                <name>removeAbandoned</name>
                <value>true</value>
        </parameter>
        <parameter>
                <name>removeAbandonedTimeout</name>
                <value>20</value>
        </parameter>
        <parameter>
                <name>logAbandoned</name>
                <value>true</value>
        </parameter>
</ResourceParams>

When you have a mis-behaving JSP (one that doesn't return its
connection), you'll get a stack trace in catalina.out (or wherever you
have redirected catalina.out) that contains the name of the JSP or class
that did not return a connection, and that was forced abandoned by the
pool. With the above config, this happens in 20 seconds (though it won't
be logged until the *next* access of the pool).  I'm not familiar with
the Oracle drivers, but hopefully they have something similar?
 
The reason I think your developers are blowing smoke... You are using
4.1.x and they are quoting 3.x docs.  They should know better!



> -----Original Message-----
> From: Stephen Carville [mailto:[EMAIL PROTECTED] 
> Sent: Monday, March 01, 2004 9:18 AM
> To: Tomcat Users List
> Subject: Re: Virtual Hosts with Apache and Tomcat
> 
> 
> Here is some more information on the problem. From a developer:
> 
> "According to the document that the link below refers to, a 
> single instance of Tomcat will have multiple JVMs, where each 
> JVM represents a virtual host.  The following link clearly 
> states this virtual host concept as it applies to Tomcat. 
> http://jakarta.apache.org/tomcat/tomcat-3.2-doc/uguide/tomcat_ug.html
> (please refer the virtual host section).}"
> 
> "As per the above document, each JVM corresponding to a 
> virtual host contains a database connection pool object. 
> Hence the connection pool that has been implemented seems to 
> be in-line with the virtual host definition in the above document.
> 


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to