Undeployment incomplete with cloud-registry-service

[rju]

When I deploy the cloud-registry-servicevia install the first time, the operation is performed successful. However, after undeployment (either via mvn or via glassfish UI) a new deployment does not work at all. I have to stop the glassfish service, go to its /opt/glassfish4.0/glassfish/domains/cocome/application directory (cocome is the domain) and actively delete cloud-registry-service directories inapplicationandapplication/__internal'. Then I must open cocome/config/domain.xml and remove all references to cloud-registry-service. After all that I can deploy again for one single time.

[tpoeppke]

I had the same problem with enterprise-logic-ear and store-logic-ear but could not reproduce it for the cloud-registry-service. The problem seems to be that the log4j library jar is kept as a stale file after undeployment if it is directly included in the webservice projects. I was able to undeploy and deploy everything three times in a row after the change in b5fc8d2.

[rju]

I have the impression that there are several different things which can go wrong in the deployment phase which the cargo and the glassfish-maven plug-in cannot handle. Therefore, I added a small script which helps deploying the stuff by hand. We should add all these different errors and solutions to a trouble shooting document (I added one to cocome-maven-projec/doc). Maybe we should move this documentation to a wiki page. For now we can add ideas to the documents there.