Glassfish deployment errors with uninformative error messages

Question

I am developing a Java EE enterprise application using NetBeans IDE. I can develop and test my whole application on the Glassfish instance bundled with NetBeans IDE, but I face kind of vague errors when I try to deploy my application on our production Glassfish server. Every time I try to deploy the .ear package on the server, Glassfish reports this error:

SEVERE: Exception while loading the app
SEVERE: Exception while shutting down application container
SEVERE: Exception while shutting down application container : java.lang.NullPointerException
SEVERE: java.lang.RuntimeException: Error occurred during deployment: Exception while shutting down application container : java.lang.NullPointerException. Please see server.log for more details.

The application is developed from scratch on NetBeans 7.2, Java EE 6, JDK 7 and deployed on GlassFish 3.1.2.2. Changing the log levels to ALL does not make GlassFish produce more helpful error message. There are other solutions mentioned on the web, including creating an empty, 0-byte beans.xml file and using another application server such as JBoss. I didn't try JBoss or other app servers, but cannot find what causes the error on Glassfish.

Answer 1

This in not an universal answer, but I faces with this problem when define session scope class without implementing Serializable interface. Example with NullPointerException:

import javax.enterprise.context.SessionScoped;
import javax.inject.Named;

@Named
@SessionScoped
public class someBean(){
}

Example without NullPointerException:

import javax.enterprise.context.SessionScoped;
import javax.inject.Named;

@Named
@SessionScoped
public class someBean() implements Serializable{
}

System Enviroment: Windows x64, jdk x64, glassfish x64.

Answer 2

If your app is fine, than usually glassfish failed to undeploy something cleanly. This of course implies that you had that particular app deployed before. What I usually do to fix this is:

• stop glassfish
• go to your domain and/or instance directory and check the following directories if anything named like your app is there and delete it: 'application', all the subfolders under 'generated'.
• delete the felix directory under osgi (make sure you don't wipe that directory in your install directory!)
• start glassfish
• deploy

In case you are running a glassfish cluster, you usually need to do this both on your DAS and your instances. Very annoying, but that usually only happens after quite a few deployments.

Answer 3

Strangely, the 32-bit JDK running on 64-bit Debian 6 system was causing the error. I have tried every possible relevant and non-relevant thing I could think of, including using the same JDK update, changing /etc/hosts to match the development system, installing NetBeans with bundled Glassfish on the production server (!), using an invalid IP address instead of a valid one, using Oracle Glassfish server instead of the community version, etc. None of them helped. The problem got fixed only after I noticed the 32-bit JDK and replaced it with a 64-bit version.

The error, its vague reporting and somehow non-relevant cause wasted 5 days of my time. I decided to share my experience, hopefully it helps someone in a similar frustrating situation.