This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.
Build: milestone build 17 all 4 TRS tests are failed on Mac OS (both Power PC & Intel). Found a bunch of exceptions in server.log file. (see attached error.log file) Sych sample and Asych sample are fine.
Created attachment 32623 [details] exceptions found in server.log file
Created attachment 32624 [details] Actual output file
Please Hong, provide much more information at this point. Given what you know about the interdependence of pieces and parts of the TRS sample. Can you provide any further information. Is this a regression? Was it ever working before? Do you know if the EJB/J2EE app deployed successfully or not? Do you get the same results on another MAC machine? Are you sure all 4 test cases behave the same way? I would expect slightly different behavior from the HasHotel test case. At least if it is run first following a server restart perhaps.
1. It is a regression. All TRS tests passed in TPR3. 2. Yes. EJB/J2EE app deployed successfully. Checked "Enterprise Application" and "JMS Resources" in Runtime window, all look ok. 3. I tried to run TRS tests on 3 Mac machines (2 Power PCs, and 1 Intel). Got the same result. 4. 1) I ran "Test Project" first, only the first test HasAirline generated Actual output file. The rest of three tests doesn't even generate any acutal output file. And they all failed. 2) Ran each test again. Still failed, without generating any output file. 3) Stopped App Server and re-started it. Ran HasHotel test first. It passed. Then the first test being run next generated an actual output file. Almost like the attached Actual_20060807165325.xml. Then no more actual output file is generated if runs the rest of tests before re-starting App Server again. Even HasHotel will fail if it is run again.
the fact that HasHotel works ok following a restart suggests that the problem is the connectio b/t BPEL SE and App Server JMS or EJB is problematic. Also suggests that once a single correlated instance is messed up then subsequent tests are in trouble. This latter fact is probably related to correlation set being the same for all test runs. One minor check to clarify the severity of server lock up would be to change the correlation set value in the HasHotel test run input message to see if it continues to generate output.xml once you use a different correlation set id. I hope so, because otherwise problem is even more severe. That test will just help us ensure that the runtime is still allowing new instances with different correlation ids to be created. It will not help us solve the actual problem but it should be undertaken to help clarify severity of the problem.
Tested in spb on Mac with slightly older build from last Thursday. No problems were encountered with deploy and test run. There was one unrelated problem noted in bug 81516. I repeat this had no impact on deploy and test run, which were successful. I only note this here because perhaps this other bug has not been fixed for Mac OS. We were on corporate network, which needs proxy server to access internet. The runtime tab was not showing any details under the JBI nodes. This seems to be related to the "JBI nodes not working propely when not connected to internet" problem which was reported earlier. I don't know bug number right now. This machine was connected to internet but it required proxy server settings. So question is whether this problem is already noted. I see that bug was fixed on Thursday so fix is possibly not in the build which was just tested. We will retest this as well as the TRS test runs on latest build.
The reason that cause the tests fail is, in order to run TRS sample application, Application Server needs to be on port 8080. Otherwise, some modifications in source codes are needed. Can find the detail in bpel getting start guide, trouble shooting session (http://www.netbeans.org/kb/55/bpel_gsg.html#troubleshooting).
Bug is invalid.