This message is for David Jencks, please read this.

[juanchi]

Hello David,

I write four o five message in this forum but never someone answer my questions.

Please I'm working in my project of the University. I need to know the administration of connection pool to de JDBC. This is very important because a study this situation of JBoss. I need write a conclusions of connections (thread) if there are a lot of clients to connect to the database.

I need to know if there are one thread for connection, if this thread are native or green, if the thread who work in JBoss with the instance of ejb is the same of the thread who connecto to the DB.

Please I need this information or what is the way that I try for know this.

Sorry for my english, but I'm Spanish.

A lot of thanks for all.

Juanchi.

[davidjencks]

I'm not at all sure I understand what all your questions are, I'll do my best.

The jca/jdbc framework in jboss does not create any threads to do work on. Connections are always used in the same thread as the ejb/servlet/jsp/mbean that uses the connection.

The number of threads in jboss is really determined by the invoker you use (for calls from a remote client to ejbs) or the servlet engine when it starts to process an http request.

For most invokers, the number of threads cannot be limited: in my opinion this is a serious weakness in jboss. As far as I know the only invoker that provides control over the thread count is the TrunkInvoker in jboss 4.

[davidjencks]

I forgot to mention, the thread type depends of course on what jvm you are running and how it is configured. JBoss has no control over this, although you can probably set it with startup command line parameters.

[juanchi]

Hi David,

First of all, thanks.

In my projecte I study the work of the threads what JBoss uses for de connections.

The conclusion is... the thread is created in the http and all the time is the same?

In JBoss there aren't a pool of thread for the DataSource?

This is correct? If this is correct... my project

The datasource haven't a pool of threads? I believed that every datasource have a pool of threads.
breackdown...ummm

[davidjencks]

why would a datasource have a pool of threads? every call to a datasource is synchronous and must return before any work in the calling thread may proceed.

The only reason I can think of to run work in a separate thread is to enable abandoning it after a timeout. I'm not sure this is a sufficient reason.

[juanchi]

Ok David.

Thanks for all.

If I've any question more, can I put other message for you with subject ..."This message is for David Jencks"?

Juanchi.