[Solved]Gemini - Maven-tycho build works on some machines, but not others. [message #968452] |
Fri, 02 November 2012 09:24  |
Eclipse User |
|
|
|
Introduction:
We deploy our Gemini system with a declarative service (within a PDE) that returns a EMF to another plugin for testing. The PU are in their own directory (a third bundle) and there is only one persistence.xml file.
I have figured out how to get Gemini working with our current testing infrastructure with a Run Configuration file and it builds (maven) perfectly fine on our remote workstation (16 core RAIL running Fedora). I also have it working on a duo core (centrino Fedora 16)) laptop and a MAC workstation.
Nevertheless, the build fails with a quad core (core 2 quad windows 7) box and a 8 core rail (Fedora) with an "Object is unknown entity type" exception. Keep in mind that the EMF and EM do have meta data for the entity types.
I have ran the Finest level of debug, and I am not seeing any real difference between the ones that pass the build and the ones that fail the build. I have weaving set to false and the refresh set to false. I was wondering if anyone has ran into this type of issue before and has some suggestions on how to fix it.
Keep in mind that our JUNIT run configuration file works perfectly on every machine. This may have something to do with incorrect run configuration in our maven build system. But I am open to any other suggestion.
I will post the current pom configuration momentarily for review.
Thoughts?
[Updated on: Mon, 05 November 2012 15:12] by Moderator
|
|
|
|
Re: Gemini - Maven-tycho build works on some machines, but not others. [message #968589 is a reply to message #968452] |
Fri, 02 November 2012 11:26  |
Eclipse User |
|
|
|
Seems I fixed it by adding my PU to the start levels (start level 3, auto start true) along with the DS plugin (after PU, start level 3). I can not believe this configuration worked on some machines but not others. I can only speculate that this erroneous event was some kind of race condition where the PU plugin was being started by some other plugin just at the right moment for some machines, but not others.
If this holds over the weekend (we have continuous builds every hour), I will mark this as solved.
[Updated on: Fri, 02 November 2012 11:27] by Moderator
|
|
|
Powered by
FUDForum. Page generated in 0.04047 seconds