We Keep Coding

EJB with CMT when migrate from JBoss 7 to WildFly 9

I'm migrating my application from JBoss 7 to WildFly (v9.0.1) and it is not deployed because of bean transaction management error.
Caused by: javax.naming.NamingException: WFLYNAM0062: Failed to lookup env/com.component.eventmgt.EventServiceImpl/transaction [Root exception is java.lang.RuntimeException: WFLYNAM0059: Resource lookup for injection failed: java:jboss/UserTransaction]
at org.jboss.as.naming.ServiceBasedNamingStore.lookup(ServiceBasedNamingStore.java:157)
at org.jboss.as.naming.ServiceBasedNamingStore.lookup(ServiceBasedNamingStore.java:83)
at org.jboss.as.naming.NamingContext.lookup(NamingContext.java:207)
at org.jboss.as.naming.NamingContext.lookup(NamingContext.java:193)
at org.jboss.as.naming.NamingContext.lookup(NamingContext.java:189)
at org.jboss.as.naming.deployment.ContextNames$BindInfo$1$1.getReference(ContextNames.java:316)
... 90 more
Caused by: java.lang.RuntimeException: WFLYNAM0059: Resource lookup for injection failed: java:jboss/UserTransaction
at org.jboss.as.naming.deployment.ContextNames$BindInfo$1$1.getReference(ContextNames.java:319)
at org.jboss.as.naming.ServiceBasedNamingStore.lookup(ServiceBasedNamingStore.java:143)
... 95 more
Caused by: javax.naming.NameNotFoundException: UserTransaction [Root exception is java.lang.IllegalStateException: WFLYEJB0137: Only session and message-driven beans with bean-managed transaction demarcation are allowed to access UserTransaction]
at org.jboss.as.naming.ServiceBasedNamingStore.lookup(ServiceBasedNamingStore.java:153)
at org.jboss.as.naming.ServiceBasedNamingStore.lookup(ServiceBasedNamingStore.java:83)
at org.jboss.as.naming.NamingContext.lookup(NamingContext.java:207)
at org.jboss.as.naming.NamingContext.lookup(NamingContext.java:193)
at org.jboss.as.naming.NamingContext.lookup(NamingContext.java:189)
at org.jboss.as.naming.deployment.ContextNames$BindInfo$1$1.getReference(ContextNames.java:316)
... 96 more
Here is the EventServiceImpl class.
#Stateless
#Remote(EventService.class)
#TransactionAttribute(TransactionAttributeType.REQUIRED)
public class EventServiceImpl implements EventService {
/**
* Logger
*/
private static Logger log = LoggerFactory.getLogger(EventService.class);
private EventTableDAO eventDao;
#PersistenceContext(unitName = "SOMF-GT")
private EntityManager entityManager;
#Resource
private UserTransaction transaction;
public List<Map> loadEvents() throws EventsException {
Configuration configurationEntry = new Configuration();
try {
Map configuration = configurationService.getConfiguration();
if (configuration != null) {
eventDao = new EventTableDAO(Event.class, entityManager, transaction);
List<Map> eventsMapList = new ArrayList();
}
}
I know that if i changed the transaction management to BMT with #TransactionManagement(TransactionManagementType.BEAN) but then the following error emerge
WFLYJPA0060: Transaction is required to perform this operation (either use a transaction or extended persistence context)
I want to know why we have to change this in the first place ?
Any information, please !

These changes rolled out in Wildfly 8 and were (as noted below) based on standardization of the global JNDI namespace in EJB 3.1.
From the Wildfly 8 Developer Guide:
EJB 3.1 introduced a standardized global JNDI namespace and a series of related namespaces that map to the various scopes of a Java EE application. The three JNDI namespaces used for portable JNDI lookups are java:global, java:module, and java:app. If you use JNDI lookups in your application, you will need to change them to follow the new standardized JNDI namespace convention.
To conform to the new portable JNDI namespace rules, you will need to review the JNDI namespace rules and modify the application code to follow these rules.
The guide further notes:
WildFly 8 has tightened up on JNDI namespace names to provide predictable and consistent rules for every name bound in the application server and to prevent future compatibility issues. This means you might run into issues with the current namespaces in your application if they don't follow the new rules.
Here's a snippet from the table showing Examples of JNDI mappings in previous releases and how they might look now specific to the UserTransaction:
Previous Namespace New Namespaces
------------------ --------------
java:comp/UserTransaction java:comp/UserTransaction (This will not be accessible for non EE threads, e.g. Threads your application directly creates)
java:comp/UserTransaction java:jboss/UserTransaction (Globally accessible, use this if java:comp/UserTransaction is not available)
Edit re: WFLYEJB0137:
This is theory-craft and may be worthless - let me know and I'll delete it. Java EE 6 Tutorial - Container-Managed Transactions says:
Enterprise beans that use container-managed transaction demarcation also must not use the javax.transaction.UserTransaction interface.
Further:
(Transaction) Required Attribute
If the client is running within a transaction and invokes the enterprise bean’s method, the method executes within the client’s transaction. If the client is not associated with a transaction, the container starts a new transaction before running the method.
The Required attribute is the implicit transaction attribute for all enterprise bean methods running with container-managed transaction demarcation. You typically do not set the Required attribute unless you need to override another transaction attribute. Because transaction attributes are declarative, you can easily change them later.
The exception message pretty much says it all:
WFLYEJB0137: Only session and message-driven beans with bean-managed transaction demarcation are allowed to access UserTransaction
Your EJB is using container managed transaction (CMT) demarcation which does not inter-operate with bean managed transaction (BMT) demarcation wherein UserTransaction lies.
Regarding a switch to BMT and
WFLYJPA0060: Transaction is required to perform this operation (either use a transaction or extended persistence context)
I found Transaction is required to perform this operation (either use a transaction or extended persistence context) which seems to indicate that the transaction is to be managed by you, as you noted in your comments #Marco. It appears that you have made the appropriate modification.

Java Batch(JSR-352) and Session Context CDI

I am using the java Batch (JSR-352), it is possible to work with a session bean inside it? I needed to have a Bean with #SessionScope annotation, to catch some information in it, that to differentiate the type of User that is running the batch process.
It's possivel use a Session Context CDI inside the specification ? If is possible how is the best pratice

In general, the session isn't going to propagate from the thread starting the job via JobOperator to the execution thread. I don't recall if this is under discussion in the CDI specification still or a settled matter, but for now you can't.

No bean named but bean is defined

I'm working on an update version of grail-oauth-plugin that support last spring-oauth
My plugin version works good and I have implemented a workin oauth2 server.
But now I want to add a custom-grant defined like this
def doWithSpring = {
myTokenGranter(MyTokenGranter)
xmlns oauth:"http://www.springframework.org/schema/security/oauth2"
oauth.'authorization-server'( /* ... many definitions here ... */){
/* ... many definitions here ... */
oauth.'custom-grant'('token-granter-ref': "myTokenGranter")
}
}
But I get an exception telling me:
org.springframework.beans.factory.NoSuchBeanDefinitionException: No bean named 'myTokenGranter'
But the bean myTokenGranter is defined as you can see. And If I remove the custom-grant definition the project starts and I can access the myTokenGranter bean.
Looking to a fullstack trace I see that the exception occur in the spring oatuh2 server bean definition parse AuthorizationServerBeanDefinitionParser.java in the line where it try to find my bean
parserContext.getRegistry().getBeanDefinition(customGranterRef);
where customGranterRef = "myTokenGranter"
so I suspect there is a bug in Spring Ouath or in Grails BeanBuilder that does not let my "myTokenGranter" to be visible in the server parser. Or making some error in grails bean definition DSL.
Thank you for your interest.
Debugging the app more deeply I have found that the problem probably is in how grails BeanBuilder work in translating namespaced spring DSL.
If I debug the point where my bean is checked (in AuthorizationServerBeanDefinitionParser.java)
at row
parserContext.getRegistry().getBeanDefinition(customGranterRef);
if I check che result of
parserContext.getRegistry().getBeanDefinitionNames()
it show me only this beans
[org.springframework.context.annotation.internalConfigurationAnnotationProcessor
org.springframework.context.annotation.internalAutowiredAnnotationProcessor
org.springframework.context.annotation.internalRequiredAnnotationProcessor
org.springframework.context.annotation.internalCommonAnnotationProcessor
org.springframework.context.annotation.internalPersistenceAnnotationProcessor
org.springframework.aop.config.internalAutoProxyCreator
org.springframework.transaction.annotation.AnnotationTransactionAttributeSource#0
org.springframework.transaction.interceptor.TransactionInterceptor#0
org.springframework.transaction.config.internalTransactionAdvisor
oauth2TokenGranter
oauth2AuthorizationCodeServices
oauth2AuthorizationRequestManager]
And not all other decleared beans...
The problem exist even if I move the ouath server declaration inside resources.xml, keeping my custom token granter bean declaration inside resources.groovy.
But the problem solves if I move the custom token bean declaration inside resources.xml.

I don't really know how the BeanBuilder DSL works, but it seems like the problem is there if there is a problem (your example works just fine in XML). Can you do it in two steps, so the bean definition for myTokenGranter is definitely available when the OAuth2 namepsace is handled?

Solved hacking Spring Security Oauth
see this commit

Spring Injection & Globally initialized objects

I have a spring injected service
class A{
List l = new ArrayList();
public m1(){
//do some additions in list
}
public m2(){
//do some new additions in list
}
}
Now because creating of objects of A, is in the hands of spring the behavior of program is not what is expected. (I expect list to be available empty always but not initialized by methods for some wired reason)
Will Spring always create only one instance of A, so that list l will keep on growing, I have configured bean as singleton in application context.
If yes, naturally I must initialize the list inside the functions m1 & m2 or callee must past the reference, and in my case callee being struts2 actions they are not singleton so this issue can be solved?
Or
Does spring provide any support in configuration to initialize member variables at every call or something else?
More generally what are best practices to have in writing services injected by spring about using member variables/ local variables for performance & efficiency.

Trying to answer following :
Does spring provide any support in configuration to initialize member variables at every call or something else?
By default Spring beans are singleton. Initialized only once and use the same object again and again.
However, if requirement changes as you asked. You need to understand scope attribute provided by the Spring.
<bean id="id" class="com.test.TP" scope="prototype">
</bean>
Refer following for detail understanding.
Scope Description
singleton
Scopes a single bean definition to a single object instance per
Spring IoC container.
prototype
Scopes a single bean definition to any number of object instances.
request
Scopes a single bean definition to the lifecycle of a single HTTP
request; that is each and every HTTP request will have its own
instance of a bean created off the back of a single bean definition.
Only valid in the context of a web-aware Spring ApplicationContext.
session
Scopes a single bean definition to the lifecycle of a HTTP Session.
Only valid in the context of a web-aware Spring ApplicationContext.
global session
Scopes a single bean definition to the lifecycle of a global HTTP
Session. Typically only valid when used in a portlet context. Only
valid in the context of a web-aware Spring ApplicationContext.
http://static.springsource.org/spring/docs/3.0.0.M3/spring-framework-reference/html/ch04s04.html
It is also possible to have user defined scope such as thread scope.

You can configure your bean as follows:-
<bean id = "serviceBeanA" class = "somepkg.A" scope="prototype">
<property name = "l">
<value>
<list>
<value>ABC</value>
.....
</list>
</value>
</property>
</bean>

I'm not sure if I understand the question but your sample class should work fine with spring. Spring will call the default constructor unless you pass in constructor args in the configuration file. As long as you have an id="..." in the bean then it will create a singleton of that class. Then, the first class that calls a.m1() will see l as being an empty list.
What may be happening is that you have multiple instances of A being created. See here about Spring singletons versus non. To quote:
Beans are defined to be deployed in one of two modes: singleton or non-singleton. (The latter is also called a prototype, although the term is used loosely as it doesn't quite fit). When a bean is a singleton, only one shared instance of the bean will be managed and all requests for beans with an id or ids matching that bean definition will result in that one specific bean instance being returned.
So if you don't have an id or name specified in your Spring configuration then you might get multiple instances of A created.
The non-singleton, prototype mode of a bean deployment results in the creation of a new bean instance every time a request for that specific bean is done. This is ideal for situations where for example each user needs an independent user object or something similar.

Stateful EJB not being passivated when the client VM is abruptly terminated

I have an stateful session bean deployed on Jboss 4.3 with the following config on jboss.xml:
<cache-config>
<idle-timeout-seconds>7200</idle-timeout-seconds>
<remove-timeout-seconds>8000</remove-timeout-seconds>
</cache-config>
IE:
The bean should be passivated after 2 hours of IDLE and removed after 2 hours and 15 minutes.
The problem is: when the client VM is abruptly terminated the bean is neither passivated nor removed.
Is there a way to avoid this?

A container may only passivate a stateful session bean instance when
the instance is not in a transaction.
A container must not passivate a stateful session bean with an
extended persistence context unless the following conditions are met:
All the entities in the persistence context are serializable.
The EntityManager is serializable.
Any interceptor classes associated with the stateful session bean must be Serializable.
The references can possibly restrict the bean from getting passivated.