-
1. Re: NPE in StandaloneClientImpl.isDeploymentNameUnique
brian.stansberry Nov 11, 2010 10:43 AM (in response to thomas.diesler)I can't see how an NPE is possible there. On the off chance it's using some old code, I'm having a quick look at the history of that method.
-
2. Re: NPE in StandaloneClientImpl.isDeploymentNameUnique
brian.stansberry Nov 11, 2010 10:49 AM (in response to brian.stansberry)Not surprising, a look at the history doesn't show a previous form of that method that would throw an NPE on that line either. Very strange.
-
3. Re: NPE in StandaloneClientImpl.isDeploymentNameUnique
brian.stansberry Nov 11, 2010 10:58 AM (in response to brian.stansberry)Carlo found it; executeForResult must be returning null, leading to a NPE from autoboxing to boolean.
-
4. Re: NPE in StandaloneClientImpl.isDeploymentNameUnique
brian.stansberry Nov 11, 2010 11:53 AM (in response to brian.stansberry)I think Carlo also found the basic cause; there's a concurrency bug in the ManagementRequest class. It has a "responseBodyHandler" and a "responseEndHandler" and assumes those will not be executed concurrently, but they can be. One effect would be a null result gets passed to the Future.
-
5. Re: NPE in StandaloneClientImpl.isDeploymentNameUnique
thomas.diesler Nov 11, 2010 3:05 PM (in response to brian.stansberry)You are a hero. Thanks for following up on this ;-)
-
6. Re: NPE in StandaloneClientImpl.isDeploymentNameUnique
brian.stansberry Nov 12, 2010 9:33 AM (in response to thomas.diesler)Thomas, I pushed a fix for this[1] upstream. Can you let me know if that makes your EC2 failures go away?
[1] http://github.com/jbossas/jboss-as/compare/73a27f3...4a7b765
-
8. Re: NPE in StandaloneClientImpl.isDeploymentNameUnique
thomas.diesler Nov 12, 2010 4:46 PM (in response to brian.stansberry)I've seen this again even after the patch
-
9. Re: NPE in StandaloneClientImpl.isDeploymentNameUnique
thomas.diesler Nov 29, 2010 3:36 AM (in response to thomas.diesler)This now seem to occur more regularly
http://jbmuc.dyndns.org:8280/hudson/job/jbosgi-remote-matrix/container=jboss70x
-
10. Re: NPE in StandaloneClientImpl.isDeploymentNameUnique
brian.stansberry Nov 30, 2010 6:09 PM (in response to thomas.diesler)Is there any client side logging maintained? One thing I'm seeing is org.jboss.as.protocol.mgmt.ManagementRequest.responseBodyHandler lets any IOException propagate back to the Connection, which will just log it as an ERROR and carry on. This could result in the NPE.
Whether or not you have logging that demonstrates this is what's happening in your hudson failures, I need to fix the above; I should be able to get a patch done tonight; at least on a branch if not upstream.
If my theory is right though, the patch won't fix your test; it will just make the real problem more visible.
-
11. Re: NPE in StandaloneClientImpl.isDeploymentNameUnique
brian.stansberry Nov 30, 2010 9:03 PM (in response to brian.stansberry)That's pushed upstream.