[jesper.pedersen]

Sounds like that the class-name is wrong, either in :resource-adapter: or in the ra.xml. Or that the wrong data is passed down - TRACE will tell you.

[simas_ch]

I can't see a wrong name. This is all correct. And if I already told you it works in JBoss AS 6.1.0

Does Ironjacamar maybe no support JCA 1.0?

[jesper.pedersen]

JBoss AS 6.1.0 is a completely different JCA container, and yes, IronJacamar does support JCA 1.0 - its by spec.

Start by looking at the TRACE for the activation, then set break points in AbstractResourceAdapterDeployer before you see the WARN.

[jesper.pedersen]

Remember to use the implementation class of the ManagedConnectionFactory as the class name in the :resource-adapter: entry...

[simas_ch]

I got it! Unfortunatly the answer of Jeff was missleading.

Correct is:

<connection-definition class-name="com.prodega.proshop.integration.spi.GenericManagedConnectionFactory"

Wrong was

<connection-definition class-name="com.prodega.proshop.integration.IODataSource"

Thanks very much for your help, Jesper!

[yangju]

Hi, Jesper, I have a similar problem and I have struggled for a few days. The same code runs fine in jboss eap6.0.

Basically, I am using Wildfly 6.0.0.Final out of the box. We have a native sql query (insert) executed through EntityManager (em). The datasources are mysql and we are using xa datasources.

However, after the server reboot, the first time this em.executeUpdate is invoked, it always fail. See the attached log.

The subsequent em.executeUpdate() is always successful until after like 10 or 15 minutes no activity, then when it is invoked again, it would fail.

I guess this is something to do with the fact that the datasource connection is staled or not "warmed up" before jca grab it? How do I prevent this from happening?

I am attaching the datasource file also.