"phantom" ObjectNotFoundException

Question

A small (one servlet) application I have been asked to debug will
          sporadically generate many instances of the following exception. The app
          server gets so busy handling these exceptions that it is effectively shut
          down.
          ####<Feb 5, 2002 11:27:37 AM EST> <Error> <RJVM> <skip> <skip-nih>
          <ExecuteThread: '39' for queue: 'default'> <> <> <000000> <Unsolicited error
          response for: '-1'>
          java.rmi.NoSuchObjectException: RemoteInvokable - id: '377'
          <<no stack trace available>>
          The servlet generally does:
          1.) calls remote stateless EJBs
          2.) does an include of some content from a remote XML server (kinda' like a
          web service).
          3.) reads files
          4.) calls xalan XSLT to do a transform
          5.) the servlet is single threaded
          Here is what I've looked at so far:
          1.) EJB references, all to stateless session, are created immediately before
          use and
          destroyed immediately after.
          2.) The InitialContext, to a tree on a remote box, is created each time it
          is needed.
          3.) There is no explicit RMI (i.e. other than EJB) being used.
          The app server details:
          1.) WLS6.0 SP2
          2.) Redhat Linux 7.2
          3.) JDK 1.3.1
          4.) 2 physical servers in cluster (first is admin and cluster1, second is
          cluster2)
          5.) fronted by Apache
          6.) remote EJB server is similar configuration
          Questions:
          1.) How do I know it is this app? I don't, there are two apps running on
          this
          cluster but the other one is similar.
          2.) Where is the exception coming from that it cannot provide a stack trace?
          3.) Should this be posted on the EJB newsgroup?
          Any hints, help or ideas are appreciated. Please reply to newsgroup.
          

Answer 1

Chris
          I think you've hit a known issue for which there is a patch, to get it open a
          support case and reference CR053957
          Thanks
          Jim
          Chris Beckey wrote:
          > A small (one servlet) application I have been asked to debug will
          > sporadically generate many instances of the following exception. The app
          > server gets so busy handling these exceptions that it is effectively shut
          > down.
          >
          > ####<Feb 5, 2002 11:27:37 AM EST> <Error> <RJVM> <skip> <skip-nih>
          > <ExecuteThread: '39' for queue: 'default'> <> <> <000000> <Unsolicited error
          > response for: '-1'>
          > java.rmi.NoSuchObjectException: RemoteInvokable - id: '377'
          > <<no stack trace available>>
          >
          > The servlet generally does:
          > 1.) calls remote stateless EJBs
          > 2.) does an include of some content from a remote XML server (kinda' like a
          > web service).
          > 3.) reads files
          > 4.) calls xalan XSLT to do a transform
          > 5.) the servlet is single threaded
          >
          > Here is what I've looked at so far:
          > 1.) EJB references, all to stateless session, are created immediately before
          > use and
          > destroyed immediately after.
          > 2.) The InitialContext, to a tree on a remote box, is created each time it
          > is needed.
          > 3.) There is no explicit RMI (i.e. other than EJB) being used.
          >
          > The app server details:
          > 1.) WLS6.0 SP2
          > 2.) Redhat Linux 7.2
          > 3.) JDK 1.3.1
          > 4.) 2 physical servers in cluster (first is admin and cluster1, second is
          > cluster2)
          > 5.) fronted by Apache
          > 6.) remote EJB server is similar configuration
          >
          > Questions:
          > 1.) How do I know it is this app? I don't, there are two apps running on
          > this
          > cluster but the other one is similar.
          > 2.) Where is the exception coming from that it cannot provide a stack trace?
          > 3.) Should this be posted on the EJB newsgroup?
          >
          > Any hints, help or ideas are appreciated. Please reply to newsgroup.
          [Reply.vcf]