"scott.stark(a)jboss.org" wrote :
| There can be interaction between multiple ManagedDeploymentCreators. For example, with
jca, the factories(the managed components) and properties are determined by the
ManagedConnectionFactoryDeploymentGroup metadata, but there are runtime ops and stats that
have to come from the ServiceMetaData of the mbeans that the jca layer currently uses as
the runtime objects implementing the factories. Those mbeans should not be showing up as
some manageable sar, while jms destinations currently are deployed as just a collection of
mbeans.
|
Ok, I mentioned this before, but without any conclusion in how this should work.
It does need a generic mechanism where the jca created managed object
is able to respond to a request (assuming it is runtime configurable)
"mydatasource".setProperty("max-pool-size", 100);
with a redirect to the MBean (or POJO).
Only the JCA Deployer knows the ObjectName/Attribute
that "mydatasource"/max-pool-size maps to.
But that ObjectName should not be a part of what the client uses or even cares about.
That's not to say it can't get help from the real ServiceDeployer to actually
handle the dispatch of the request.
It's only the real ServiceDeployer that knows how the ServiceMetaData
really maps to the runtime objects, i.e. mbeanserver.setAttribute(...);
But that solution seems like a very long winded and brittle mechanism?
My preferred solution would be to use KernelBus, but that is not
up-to-date in terms of dispatching MBean/POJO invocations
(Attribute and InvokeDispatchContext).
View the original post :
http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4066245#...
Reply to the post :
http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&a...