4:17 a.m.
Hi all,
Re:
http://stackoverflow.com/questions/26110918/infinispan-marshalling-error-...
Re: https://github.com/hibernate/hibernate-orm/pull/820
Seems like the issue is resolved by making sure Session Factories have the same name.
Is this something new? I’ve never had the need to set it in order to get Hibernate 2LC
clustered tests working. None of my unit tests set the session factory name. One of my
most recent demos uses JPA, and never had to set hibernate.session_factory_name property.
Or is this due to the peculiarity of the Entity the user has? I’ve run some tests locally
and SessionFactoryImpl.readResolve() does not seem to get called.
@Paul/@Scott, have you seen anything similar before?
Cheers,
--
Galder Zamarreño
galder(a)redhat.com
twitter.com/galderz
6:40 a.m.
It would depend on what is inside the CacheKey, which is the key object
Hibernate passes to the second-level cache. Usually this situation comes
up with the identifier Type(s), especially in the case of composite
identifiers. Some of the Type implementations hold reference to the
SessionFactory. Serializing those properly depends on being able to
resolve the proper SessionFactory reference on deser, which can happen
based on UUID (same JVM) or name (across JVMs).
On Mon, Nov 10, 2014 at 4:17 AM, Galder Zamarreño <galder(a)redhat.com> wrote:
Hi all,
Re:
http://stackoverflow.com/questions/26110918/infinispan-marshalling-error-...
Re: https://github.com/hibernate/hibernate-orm/pull/820
Seems like the issue is resolved by making sure Session Factories have the
same name.
Is this something new? I’ve never had the need to set it in order to get
Hibernate 2LC clustered tests working. None of my unit tests set the
session factory name. One of my most recent demos uses JPA, and never had
to set hibernate.session_factory_name property.
Or is this due to the peculiarity of the Entity the user has? I’ve run
some tests locally and SessionFactoryImpl.readResolve() does not seem to
get called.
@Paul/@Scott, have you seen anything similar before?
Cheers,
--
Galder Zamarreño
galder(a)redhat.com
twitter.com/galderz
_______________________________________________
hibernate-dev mailing list
hibernate-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/hibernate-dev
10:07 a.m.
On 11/10/2014 07:40 AM, Steve Ebersole wrote:
It would depend on what is inside the CacheKey, which is the key
object
Hibernate passes to the second-level cache. Usually this situation
comes up with the identifier Type(s), especially in the case of
composite identifiers. Some of the Type implementations hold reference
to the SessionFactory. Serializing those properly depends on being able
to resolve the proper SessionFactory reference on deser, which can
happen based on UUID (same JVM) or name (across JVMs).
On Mon, Nov 10, 2014 at 4:17 AM, Galder Zamarreño <galder(a)redhat.com
<mailto:galder@redhat.com>> wrote:
Hi all,
Re:
http://stackoverflow.com/questions/26110918/infinispan-marshalling-error-...
This user found the "hibernate.session_factory_name" +
"hibernate.session_factory_name_is_jndi" settings and HHH-6822.
Re: https://github.com/hibernate/hibernate-orm/pull/820
Seems like the issue is resolved by making sure Session Factories
have the same name.
Is this something new? I’ve never had the need to set it in order to
get Hibernate 2LC clustered tests working. None of my unit tests set
the session factory name. One of my most recent demos uses JPA, and
never had to set hibernate.session_factory_name property.
Or is this due to the peculiarity of the Entity the user has? I’ve
run some tests locally and SessionFactoryImpl.readResolve() does not
seem to get called.
If the local test serializes the SF UUID and then deserializes from the
same JVM, the test should pass (as long as the same Hibernate
classloader is used). If you are involving multiple JVMs, the SF UUIDs
are local to each JVM, so you should see failures when deserializing on
a different JVM (as the SF UUID lookup will fail).
@Paul/@Scott, have you seen anything similar before?
For EE container deployments, we set the
"hibernate.session_factory_name" if the application doesn't. We set it
to an application scoped unique name (deployment archive name (including
sub-deployment reference) and the persistence unit name).
Cheers,
--
Galder Zamarreño
galder(a)redhat.com <mailto:galder@redhat.com>
twitter.com/galderz <http://twitter.com/galderz>
_______________________________________________
hibernate-dev mailing list
hibernate-dev(a)lists.jboss.org <mailto:hibernate-dev@lists.jboss.org>
https://lists.jboss.org/mailman/listinfo/hibernate-dev
3717
days inactive
3717
days old
2 comments
3 participants
participants (3)
-
Galder Zamarreño -
Scott Marlow -
Steve Ebersole