1 2 Previous Next 17 Replies Latest reply on Jun 6, 2010 6:48 PM by unwired Go to original post
      • 15. Re: potential quartz/threading or esb bug
        unwired

        Hi,

         

        I tested out the latest version off CAMEL-2784 and it seems the deploy problem may still exist.  The logs are very similar (see attached).

         

        I diffed the original and revised jars and only the QuartzComponent class differed so hopefully i'm on the right version.

         

        thanks heaps for the investigation

         

        cheers

        • 16. Re: potential quartz/threading or esb bug
          davsclaus

          Thanks for testing again.

           

          Its definitely an interesting perspective in terms of deployment on OSGi. I really wonder if it would be much simpler on OSGi if it had a queue of deployment tasks and executed them serialized. Or was able to detect deployments which involves shared bundles and thus will be able to queue those deployment, so they are executed in serial order.

           

          What happens from the logs is that one is shutting down, while another is starting, and they get interleaved.

          • 17. Re: potential quartz/threading or esb bug
            unwired

            Thanks for the feedback.

             

            Your explanation is interesting because we are actually experiencing issues on restarts of the server, even aside from the quartz issues we've raised.  I agree that a serialized execution should sort this out; I can only imagine asynchronous loading would be used to try to save time bc osgi purports a fast startup.

             

            Also, just a little feedback  - in the wrong forum now as this is for servicemix -, but we also see issues simply deploying bundles. It  is not uncommon at all to deploy a bundle, get an Invalid Bundle Context exception, which then seems to put the whole server in a bad "state".  We then have to bounce the whole server - which prior to your solution for the quartz issue was a major pain.. but now at least we can do an easier restart.

             

            We are planning to trial spring's dm server when we have a little time to see if some of these issues aren't resolved. I wonder if they do synchronous loading; I'm sure that's how they load spring beans but not sure if their osgi server follows suit.

             

            Thanks for the feedback anyway. We'll plunge forward with your latest fix.

             

            thanks for the help/feedback!

             

            Edited by: unwired on Jun 6, 2010 10:47 PM

            1 2 Previous Next