Carlos Sierra commented on CXF-7409:

In our case this address is coming from a property on a OSGi service reference. 

We are avoiding it at the moment using a custom destination Factory to detect 
the conflict before it gets to destroy the server. 
To be honest I did not think having several applications per servlet would be a 
problem. Would you suggest we spawn a servlet for each of the applications? If 
you think CXF does not support other way I will refactor and make sure there is 
only one application per servlet (if no other considerations apply)

I will try and find some time to create a test for you. I need to get 
acquainted with your tree. But with you comments I guess you encouraged me to 
change our implementation and completely work around this problem from the 

> ServiceConstructionException when adding JAX-RS application ruins existing 
> applications
> ---------------------------------------------------------------------------------------
>                 Key: CXF-7409
>                 URL: https://issues.apache.org/jira/browse/CXF-7409
>             Project: CXF
>          Issue Type: Bug
>          Components: JAX-RS
>    Affects Versions: 3.1.10
>            Reporter: Carlos Sierra
> if a new application conflicts with
> an existing one, an Exception:
> ---
> Caused by: org.apache.cxf.service.factory.ServiceConstructionException:
> There is an endpoint already running on /test-application.
>       at
> org.apache.cxf.jaxrs.JAXRSBindingFactory.addListener(JAXRSBindingFactory.java:86)
>       at org.apache.cxf.endpoint.ServerImpl.start(ServerImpl.java:123)
>       at
> org.apache.cxf.jaxrs.JAXRSServerFactoryBean.create(JAXRSServerFactoryBean.java:206)
> ---
> is raised, which is expected. The problem is that the already existing
> application ceases to work in the bus.
> Following [~sergeyb]'s advice I tried removing 
> {code:java}
> server.destroy()
> {code}
> from the 
> {code:java}
> catch(RuntimeException e) {}
> {code}
> block in JAXRSServerFactoryBean's create method. However there is a comment 
> saying that that invocation is there to prevent leaks. Also, invocations to 
> server.destroy() in other moments don't affect other existing applications in 
> the bus. 

This message was sent by Atlassian JIRA

Reply via email to