JDBC Connector : pool gets depleted

Question

Hi,
we are currently working with SAP EP 6.0 SP2 and trying out some development work with the connector framework.
We are writing a component which uses the CF to retrieve information from a Microsoft SQL Server 2000.
The portal component works, we are able to retrieve the information from the DB, but after some time the request times out.
We monitored the connections for a while both on the J2EE platform and on the backend database and came to this conclusion : apparently each request - even by the very same user looking at the very same component, just but pressing the refresh button in the browser window - creates a new connection in the pool managed by the connector framework.
- result : the number of successful views of the components is equal to the number of Maximum Connections defined for the JDBC connector on the J2EE platform.
In our code we are using the close() method on the IConnection interface as indicated in the documentation that comes with the PDK.
Is this a known problem OR are we doing something wrong here ?
We suspect that there is something wrong in the implementation of the JDBC connector, which isn't handling the close method correctly.

Answer 1

In more general terms this issue raises the following questions :
- how does the Connector framework manage the pool of JDBC connections ?
- the fact that our component at each request (even after a refresh in the browser) causes  the CF to create a new connection, does that mean we are writing our code incorrectly so that each request is indeed considered to be a unique and in need of its own database connection ?
We have tried to avoid as much reason as possible for the CF to generate a new connection : e.g. we have set up user mapping so that all users access the database with the same account. This should already eliminate the requirement to have a separate for each user. => in general : what are the different ways by which to control the generation of connections by the Connector Framework, for the JDBC Connector in particular ?
- So far we are writing the code accessing the SQL server straight into a JSPDynpage. Am I wrong in thinking that this is allowed. Although we intend to write a portal service as a wrapper for all database interaction in a later stage, for the simple exercises now we are directly using the CF service API in the portal components.
- by what algorithms does CF determine whether to take a connection from the pool or to create a new connection to the back-end system.
- Am I wrong in thinking that connection pool management is no longer the task of the developer when using CF ?