[hibernate-dev] Documentation on OGM + JBoss AS

Scott Marlow smarlow at redhat.com
Wed Nov 14 14:00:06 EST 2012

Adding the "services=import" to the Hibernate ORM module helped us get 

We are making it to org.hibernate.ogm.persister.OgmEntityPersister but 
get stuck on Infnispan 5.2.x not liking 

http://pastie.org/5378295 is from AS master which includes Infinispan 

I added my module definitions here 

On 11/14/2012 12:42 PM, Scott Marlow wrote:
> On 11/14/2012 12:01 PM, Emmanuel Bernard wrote:
>> 1. is the one that must be solved because if the integrator is not seen
>> then many other things won't work in Hibernate OGM, we heavily
>> reconfigure components.
> Okay, good to know that this is what we need to fix.  I'll try again to
> get this to work by changing the OGM/ORM module definitions.  I might of
> missed the "services=import" part in my last attempt.
>> 2. is not a real problem, it's just that the configuration logic would
>> need to be a bit different but AFAIR that was not trivial to do.
>> I'm happy with the placeholder
>> Is there a way for Hibernate OGM to detect that the integrator has not
>> been seen so that we can raise a clearer exception? I don't think so but
>> we can open a JIRA to investigate further.
>> On Wed 2012-11-14 11:05, Scott Marlow wrote:
>>> Which failure do we want to address?
>>> 1.  The integrator approach is not helping when using OGM/ORM on AS7.
>>> 2.  The earlier failure that I haven't yet seen when the place
>>> holder is removed and we bootstrap with
>>> HibernateOgmPersistence.createContainerEntityManagerFactory().
>>> Something else?
>>> On 11/13/2012 09:20 AM, Emmanuel Bernard wrote:
>>>> If you remove the place holder it will fail as well. Earlier actually,
>>>> so no.
>>>> On Tue 2012-11-13  8:20, Scott Marlow wrote:
>>>>> Let me know if you want me to try changing the OGM code that sets
>>>>> "---PlaceHolderDSForOGM---" to only set the datasource if its not
>>>>> already specified.
>>>>> For resolving why the OGM integrator being on the ORM classpath, is
>>>>> not enough to avoid the ORM datasource lookup, I'm not sure.
>>>>> On 11/12/2012 01:43 PM, Scott Marlow wrote:
>>>>>> Hardy pointed me to
>>>>>> https://github.com/hferentschik/ogm-kitchensink/blob/master/.openshift/config/modules/org/hibernate/ogm/module.xml
>>>>>> which duplicates some other AS modules jars in the OGM module.  That is
>>>>>> going to be specific to which version of AS is being targeted (IMO) and
>>>>>> not something I recommend that we depend on.
>>>>>> On 11/12/2012 12:46 PM, Emmanuel Bernard wrote:
>>>>>>> Hum, yes make sure to check the pom. There has been a few changes to
>>>>>>> some superclasses we extend. I did some changes in master for that.
>>>>>> I was able to build OGM master against ORM 4.1.6 without getting build
>>>>>> errors.
>>>>>>> Emmanuel
>>>>>>> On Mon 2012-11-12 12:00, Scott Marlow wrote:
>>>>>>>> Does it matter which version of Hibernate ORM is used (4.1.6 is what
>>>>>>>> we are currently using on AS master branch)?
>>>>>>>> Which version of OGM should I try to use?  Previously, I built OGM
>>>>>>> >from git master but I noticed that there is a beta1 available also.
>>>>>>>> On 11/10/2012 05:41 PM, Scott Marlow wrote:
>>>>>>>>> Hi Sanne,
>>>>>>>>> I'm not sure why the module names didn't show for ogm/orm.
>>>>>>>>> I'll try again with the demo app.
>>>>>>>>> Scott
>>>>>>>>> Sent from my tardis
>>>>>>>>> -----Original Message-----
>>>>>>>>> From: Sanne Grinovero [sanne at hibernate.org]
>>>>>>>>> Received: Saturday, 10 Nov 2012, 2:56pm
>>>>>>>>> To: Scott Marlow [smarlow at redhat.com]
>>>>>>>>> CC: Emmanuel Bernard [emmanuel at hibernate.org]; Hibernate Dev [hibernate-dev at lists.jboss.org]
>>>>>>>>> Subject: Re: [hibernate-dev] Documentation on OGM + JBoss AS
>>>>>>>>> Hi Scott,
>>>>>>>>> we don't have that error when testing on AS 7.1.1 with the
>>>>>>>>> kitchensink-ogm demo [1] so I'm confident the error is caused by some
>>>>>>>>> difference in the module; indeed I remember it being tricky to put the
>>>>>>>>> right things together: I can't test it on 7.2-SNAPSHOT now.
>>>>>>>>> We definitely don't specify any datasource in the demo configuration,
>>>>>>>>> but a custom module is required.
>>>>>>>>> Side question on the stacktrace you posted: why is the version number
>>>>>>>>> missing in the stack for Hibernate packages? Do we need to add that in
>>>>>>>>> the manifest?
>>>>>>>>> Sanne
>>>>>>>>> 1 - latest version at https://github.com/hferentschik/ogm-kitchensink
>>>>>>>>> On 9 November 2012 20:08, Scott Marlow <smarlow at redhat.com> wrote:
>>>>>>>>>> On 11/09/2012 11:37 AM, Emmanuel Bernard wrote:
>>>>>>>>>>> On Fri 2012-11-09 11:15, Scott Marlow wrote:
>>>>>>>>>>>> On 10/01/2012 11:12 AM, Emmanuel Bernard wrote:
>>>>>>>>>>>>> On Mon 2012-10-01 10:51, Scott Marlow wrote:
>>>>>>>>>>>>>> I got further yesterday (after removing the
>>>>>>>>>>>>>> ---PlaceHolderDSForOGM--- property from HibernateOgmPersistence).
>>>>>>>>>>>>>> Not sure what the long term change is (perhaps we should only add
>>>>>>>>>>>>>> the ---PlaceHolderDSForOGM--- for the non-container case).
>>>>>>>>>>>>> That may have changed but this place holder was mandatory to get OGM
>>>>>>>>>>>>> to start in some environment.
>>>>>>>>>>>>> At least make sure that these change does not break the SE mode.
>>>>>>>>>>>> People are still hitting the "Caused by:
>>>>>>>>>>>> org.hibernate.service.jndi.JndiException: Unable to lookup JNDI name
>>>>>>>>>>>> [---PlaceHolderDSForOGM---]" error.
>>>>>>>>>>>> Since people are still hitting the ---PlaceHolderDSForOGM--- induced
>>>>>>>>>>>> failure (on AS7), we should create a jira for addressing it.  The
>>>>>>>>>>>> fix could be to only add ---PlaceHolderDSForOGM--- when booting in
>>>>>>>>>>>> standalone mode but not container.
>>>>>>>>>>> The placeholder is there because otherwise, some part of Hibernate
>>>>>>>>>>> EntityManager's configuration bootstrap complain.
>>>>>>>>>>> But this value should never be used because Hibernate OGM has an
>>>>>>>>>>> integrator that overrides the ConnectionProviderInitiator with a Noop
>>>>>>>>>>> version.
>>>>>>>>>>> The fact that Sergio's stacktrace shows
>>>>>>>>>>> Caused by: org.hibernate.service.jndi.JndiException: Unable to lookup JNDI name [---PlaceHolderDSForOGM---]
>>>>>>>>>>>             at org.hibernate.service.jndi.internal.JndiServiceImpl.locate(JndiServiceImpl.java:68)
>>>>>>>>>>>             at org.hibernate.service.jdbc.connections.internal.DatasourceConnectionProviderImpl.configure(DatasourceConnectionProviderImpl.java:116)
>>>>>>>>>>> is proof to me that the OgmIntegrator is not taken into account which
>>>>>>>>>>> means that the file
>>>>>>>>>>> META-INF/services/org.hibernate.integrator.spi.Integrator
>>>>>>>>>>> is not seen by Hibernate ORM when it bootstraps the application which
>>>>>>>>>>> means that Hibernate OGM's classpath is not visible where it should be.
>>>>>>>>>> I tried adding the Hibernate OGM module to the ORM module classpath and
>>>>>>>>>> still get the same error (http://pastie.org/5352457).
>>>>>>>>>>> That's why I strongly suspects that this is either a JBoss modules bug
>>>>>>>>>>> or some misconfiguration of the ogm modules.
>>>>>>>>>>> We could try and give a better error report, open an issue for that but
>>>>>>>>>>> at the end of the day, something is broken in the ORM-AS integration wrt
>>>>>>>>>>> integrator and service discovery.
>>>>>>>>>>> Emmanuel
>>>>>>>>>> _______________________________________________
>>>>>>>>>> hibernate-dev mailing list
>>>>>>>>>> hibernate-dev at lists.jboss.org
>>>>>>>>>> https://lists.jboss.org/mailman/listinfo/hibernate-dev
>>>>>> _______________________________________________
>>>>>> hibernate-dev mailing list
>>>>>> hibernate-dev at lists.jboss.org
>>>>>> https://lists.jboss.org/mailman/listinfo/hibernate-dev
> _______________________________________________
> hibernate-dev mailing list
> hibernate-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/hibernate-dev

More information about the hibernate-dev mailing list