users@genericjmsra.java.net

[Generic JMSRA-Users] Re: WebSphere Liberty Profile Beta

From: Nigel Deakin <nigel.deakin_at_oracle.com>
Date: Thu, 07 Aug 2014 11:39:26 +0100

Larry,

Once you have got this all working (and assuming it does indeed work to your satisfaction) it would indeed be very
helpful if you could write up what you have learned. We could either add it to genericjmsra.java.net or add a link to an
article hosted elsewhere.

Thanks,

Nigel

On 06/08/2014 21:49, Larry McQueary wrote:
> Nigel,
>
> Very kind of you to respond. I hear you might know a thing or two about JMS! ;-)
>
> I'm happy to report that the source of my issues was chiefly, as you suggest, in getting the WebSphere-specific bindings
> right. IBM's documentation is formidable, in both positive and negative ways. It lays out quite a bit of information
> about both the standard bindings and non-standard bindings, but doesn't give a lot of help understanding in which cases
> one applies vs. the other. Further, much of WebSphere's JMS JCA and MDB related documentation assumes you're using
> either MQ or WebSphere's internal messaging (with even more non-standard binding/notation!), therefore the examples
> require some translation as well.
>
> Once I sorted out the fact that <jmsActivationSpec> is the exclusive way to specify the binding parameters in the
> server.xml, and sorted out the non-standard notation in ibm-ejb-jar-bnd.xml, I was able to drive my request-reply MDB
> via GenericJMSRA (and EMS) without issue.
>
> As the steps for configuring for WebSphere Liberty Profile are a bit different from those for JBoss, SJSAS/GlassFish,
> and even WebSphere "Full" Profile, I will submit my configuration steps in a separate posting so they can (I hope) be
> included in the site and/or SVN.
>
> Thanks again for the thoughtful response.
>
> Cheers,
> Larry
>
>
>
> On Wed, Aug 6, 2014 at 5:24 AM, Nigel Deakin <nigel.deakin_at_oracle.com <mailto:nigel.deakin_at_oracle.com>> wrote:
>
> Larry,
>
> I don't have any experience of the combination of app server and JMS provider you are trying to use. However
>
>
> On 31/07/2014 02:03, mcqueary_at_tibco.com <mailto:mcqueary_at_tibco.com> wrote:
>
> I am using the genericra.rar package as provided, with no modifications
> (other than adding my provider jars to it), and attempting to configure
> via the MessageDriven/__ActivationConfigProperty annotations in my MDB,
> along with the resourceAdapter and activationSpec elements of my
> server.xml. I can't seem to concoct the right activationSpec, or
> possibly my resourceAdapter element/properties are wrong. I have tried
> lots of things I see in the IBM documentation, but generally my results
> are that a) I enter a property in the activationSpec that the server
> doesn't understand (resulting in the activation spec being ignored/),
> or I enter an "empty" property, resulting in a
> NullPointerException/invalid property exception.
>
>
> The EJB spec defines that you can set activation properties either using <activation-config> elements in ejb-jar.xml
> or using @MessageDriven(__activationConfig = ... on the bean class. This should work for any app server.
>
> The app server should pass these straight through to the resource adapter.
>
> Note however that the way you actually specify that a given MDB should use a particular resource adapter is always
> non-standard and will be in some IBM-specific descriptor. This will be in the WebSphere docs. It looks as if you
> have got this part working. If you have you should see GenericJMSRA startup messages in the log when you start your
> application (and GenericJMSRA error messages if it fails).
>
> An IBM-specific descriptor may also provide an alternative, non-standard way to specify activation properties. I
> think <jmsActivationSpec> is an example of this. Again this will be in the WebSphere docs, and would use some other
> file than ejb-jar.xml. I didn't see mention of <activationSpec>.
>
> Obviously, once you have worked out how to pass activation properties to GenericJMSRA the next step is to work out
> what they should be set to...
>
> Hope this is of some limited help.
>
> Nigel
>
>