The EntityManager is injected into the test case like so
@Inject @Database(Instance.APP) private EntityManager entityManager;
The database producer creates entity managers for 2 database instances (APP
and VAULT). I originally had the getAppEntityManager() method annotated
with TransactionScoped but removing it made no difference to the flushing
problem. All other test cases which use the entity manager work fine, but
they are rather simple, ie no reading back from the database before
commiting.
Here is the producer for the EntityManager(s).
----
@ApplicationScoped
public class DatabaseProducer {
// ~---- Static Variables and Methods
------------------------------------------
private static Logger LOGGER =
LoggerFactory.getLogger(DatabaseProducer.class);
// ~---- Instance Variables
----------------------------------------------------
@Inject @PersistenceUnitName("app") private EntityManagerFactory
appEntityManagerFactory;
@Inject @PersistenceUnitName("vault") private EntityManagerFactory
vaultEntityManagerFactory;
private EntityManager appEntityManager;
private EntityManager vaultEntityManager;
private DataSource appDataSource;
private DataSource vaultDataSource;
// ~---- Public
Interface-----------------------------------------------------
@Produces @PicketLink
private EntityManager getSecuriyEntityManager() {
return getAppEntityManager();
}
@Produces @Database(Instance.APP)
private EntityManager getAppEntityManager() {
final EntityManager em = getEntityManager(Instance.APP, false);
em.setFlushMode(FlushModeType.AUTO);
return em;
}
@Produces @Database(Instance.VAULT)
private EntityManager getVaultEntityManager() {
return getEntityManager(Instance.VAULT, false);
}
private EntityManager getEntityManager(final Instance instance, final
boolean transactionScoped) {
EntityManager returnValue = null;
switch(instance) {
case APP:
if (transactionScoped) {
returnValue = appEntityManagerFactory.createEntityManager();
} else {
if (appEntityManager == null) {
appEntityManager =
appEntityManagerFactory.createEntityManager();
logDatabaseConnectionInfo(instance);
}
returnValue = appEntityManager;
}
break;
case VAULT:
if (transactionScoped) {
returnValue =
vaultEntityManagerFactory.createEntityManager();
} else {
if (vaultEntityManager == null) {
vaultEntityManager =
vaultEntityManagerFactory.createEntityManager();
logDatabaseConnectionInfo(instance);
}
returnValue = vaultEntityManager;
}
break;
}
return returnValue;
}
----
Regards,
Paul
On 14 December 2015 at 16:42, Mark Struberg <[email protected]> wrote:
> And how is the entity manager being created?
>
> LieGrue,
> Strub
>
> > Am 14.12.2015 um 04:25 schrieb John D. Ament <[email protected]>:
> >
> > Paul,
> >
> > Could you show us what your test looks like, including how it injects the
> > bean that is used here.
> >
> > John
> >
> > On Sun, Dec 13, 2015 at 10:09 PM Paul Wills <[email protected]>
> > wrote:
> >
> >> I can confirm the Transactional annotation is being imported as follows
> >>
> >> *import* org.apache.deltaspike.jpa.api.transaction.Transactional;
> >>
> >> Regards,
> >>
> >> Paul
> >>
> >>
> >>> On 10 December 2015 at 15:06, Paul Wills <[email protected]>
> wrote:
> >>>
> >>> Whilst the following method is called in a test case, run with
> >>> CdiTestRunner, the entityManager.flush() call causes a
> >>> TransactionRequiredException.
> >>>
> >>> @Transactional
> >>> public void saveAndFetchUser() {
> >>> // Given - a new user
> >>> final User randomUser = createRandomUser(false);
> >>>
> >>> // When - saved
> >>> final User savedUser = userService.saveUser(randomUser);
> >>> // Then - user is saved as well as history and can be retrieved by
> id
> >>> and username
> >>> assertThat(savedUser.getRoles(), hasSize(greaterThan(0)));
> >>>
> >>> entityManager.flush(); // manually flush as UaiCriteriaQueries
> >> aren't
> >>> flushed automatically
> >>> final Paged<UserHistory> pagedHistory =
> >>> findUserHistoryMostRecentFirst(randomUser.getUsername());
> >>> assertThat(pagedHistory.getDerivedTotal(), equalTo(1));
> >>> assertThat(pagedHistory.getList(), hasSize(1));
> >>>
> >>> ...
> >>> }
> >>>
> >>> Is this a bug, or is there another way to flush the changes to the
> >>> database?
> >>>
> >>> Environment:
> >>> weld: 2.3.1.Final
> >>> deltaspike: 1.5.1.Final
> >>> deltaspike dependencies
> >>>
> >>> <dependency>
> >>>
> >>> <groupId>org.apache.deltaspike.modules</groupId>
> >>>
> >>> <artifactId>*deltaspike*-*jpa*-module-*api*</artifactId>
> >>>
> >>> <version>${deltaspike.version}</version>
> >>>
> >>> <scope>compile</scope>
> >>>
> >>> </dependency>
> >>>
> >>> <dependency>
> >>>
> >>> <groupId>org.apache.deltaspike.modules</groupId>
> >>>
> >>> <artifactId>*deltaspike*-*jpa*-module-*impl*</artifactId>
> >>>
> >>> <version>${deltaspike.version}</version>
> >>>
> >>> <scope>compile</scope>
> >>>
> >>> </dependency>
> >>>
> >>> <dependency>
> >>>
> >>> <groupId>org.apache.deltaspike.core</groupId>
> >>>
> >>> <artifactId>*deltaspike*-core-*api*</artifactId>
> >>>
> >>> *<version>${deltaspike.version}</version>*
> >>>
> >>> <scope>compile</scope>
> >>>
> >>> </dependency>
> >>>
> >>> <dependency>
> >>>
> >>> <groupId>org.apache.deltaspike.core</groupId>
> >>>
> >>> <artifactId>*deltaspike*-core-*impl*</artifactId>
> >>>
> >>> *<version>${deltaspike.version}</version>*
> >>>
> >>> <scope>compile</scope>
> >>>
> >>> </dependency>
> >>>
> >>> <dependency>
> >>>
> >>> <groupId>org.apache.deltaspike.cdictrl</groupId>
> >>>
> >>> <artifactId>*deltaspike*-*cdictrl*-weld</artifactId>
> >>>
> >>> <version>${deltaspike.version}</version>
> >>>
> >>> <scope>test</scope>
> >>>
> >>> </dependency>
> >>>
> >>> <dependency>
> >>>
> >>> <groupId>org.jboss.weld.se</groupId>
> >>>
> >>> <artifactId>weld-*se*-core</artifactId>
> >>>
> >>> *<version>${weld.version}</version>*
> >>>
> >>> <scope>test</scope>
> >>>
> >>> </dependency>
> >>>
> >>> Regards,
> >>> Paul Wills
> >>
>
