Firstly, I am a huge fan of Wildfly.
But I reckon places where logging (or feedback in terms of Exceptions) can be improved slightly.
Take for example the process of connecting a remote EJB client to an EJB running in Wildfly 8.1.0.Final. I recently decided to try that. But it wasn't super easy, actually. I got this exception:
11:17:28,558 INFO [HelloClient] Application started 11:17:28,567 INFO [HelloClient] Creating JNDI context. For url=http-remoting://localhost:8080 [user=hello] 11:17:28,792 INFO [HelloClient] Finding EJB. JNDI name=hello-ejb-1.0-SNAPSHOT/HelloBean!hello.HelloRemote 11:17:29,233 INFO [HelloClient] Invoking sayHello() on EJB Exception in thread "main" java.lang.IllegalStateException: EJBCLIENT000025: No EJB receiver available for handling [appName:, moduleName:hello-ejb-1.0-SNAPSHOT, distinctName:] combination for invocation context org.jboss.ejb.client.EJBClientInvocationContext@4bf6ba9 at org.jboss.ejb.client.EJBClientContext.requireEJBReceiver(EJBClientContext.java:749) at org.jboss.ejb.client.ReceiverInterceptor.handleInvocation(ReceiverInterceptor.java:116) at org.jboss.ejb.client.EJBClientInvocationContext.sendRequest(EJBClientInvocationContext.java:186) at org.jboss.ejb.client.EJBInvocationHandler.sendRequestWithPossibleRetries(EJBInvocationHandler.java:253) at org.jboss.ejb.client.EJBInvocationHandler.doInvoke(EJBInvocationHandler.java:198) at org.jboss.ejb.client.EJBInvocationHandler.doInvoke(EJBInvocationHandler.java:181) at org.jboss.ejb.client.EJBInvocationHandler.invoke(EJBInvocationHandler.java:144) at com.sun.proxy.$Proxy0.sayHello(Unknown Source) at hello.HelloClient.invokeEjbMethod(HelloClient.java:72) at hello.HelloClient.main(HelloClient.java:23)
On the server - there wasn't any ERROR reporting at all (nor WARNINGS). Perhaps a TRACE line could have helped me a bit. But play a long with me...
It turned out that I forgot to add this magic property to my InitialContext bootstrapping: jboss.naming.client.ejb.context=true
And suddenly it worked.
So there could a few focus areas here to improve the situation, for example:
- Given the 'sayHello' invocation through a dynamic proxy - perhaps the client runtime could infer that the above configuration (client.ejb.context) would be implied ? Perhaps it could just state that, in the context of the EJB invocation, the property was missing?
- Let the server side callstack do some logging. That would be really appropriate. Surely it must send back a reply to the client-stack telling it that there wasn't "any EJB receiver" - so perhaps it could dump an error. I reckon that the error was that of the clients (as she didn't send the correct request) - but still, everything apart: it would be nice to know why the server sends this answer to the client runtime.
So the above is just a subject of "discussion". Have you ever had similar problems connecting remote EJB clients to @Remote EJBs in wildfly?