Rafal wrote: My advice is to set ode-axis2.threads.pool.size=1
is ode-axis2.properties. Then retry your test. Rafal, thanks for the suggestion, but trying this with Oracle DB, still produces fatal errors: FATAL - GeronimoLog.fatal(116) | Engine requested response while the message exchange hqejbhcnphr423h3vathv3 was in the state COMPLETED_OK ERROR - GeronimoLog.error(104) | The message exchange seems to be in an unconsistent state, you're probably missing a reply on a request/response interaction. DEBUG - GeronimoLog.debug(66) | Connection reclaimed by garbage collector, hostConfig=HostConfiguration[host=http://localhost:8081] DEBUG - GeronimoLog.debug(66) | Notifying no-one, there are no waiting threads Does this fatal error have any truly fatal side effects? Process execution seems to continue... > -----Original Message----- > From: Andi Abes [mailto:[email protected]] > Sent: Friday, February 20, 2009 10:59 AM > To: Alex Boisvert; [email protected] > Subject: RE: Lock exceptions > > Is this a bug for the JPA persistence then? > > > > ________________________________ > > From: Alex Boisvert [mailto:[email protected]] > Sent: Thursday, February 19, 2009 5:23 PM > To: [email protected] > Cc: Andi Abes > Subject: Re: Lock exceptions > > > > On Thu, Feb 19, 2009 at 12:46 PM, Andi Abes <[email protected]> wrote: > > Based on Alex's advice, I've replaced the derby DB, and made sure to > specify (from ODE-507) > openjpa.jdbc.TransactionIsolation= read-committed > > But I'm still not quite out of the woods: > FATAL - GeronimoLog.fatal(116) | Engine requested response while the > message exchange hqejbhcnphr423h3vathv3 was in the state COMPLETED_OK > ERROR - GeronimoLog.error(104) | The message exchange seems to be in an > unconsistent state, you're probably missing a reply on a > request/response interaction. > INFO - GeronimoLog.info(79) | ActivityRecovery: Registering activity 43, > failure reason: org.apache.ode.bpel.iapi.BpelEngineException: Engine > requested response > while the message exchange hqejbhcnphr423h3vathv3 was in the state > COMPLETED_OK on channel 75 > > > This does not happen if there's just 1 branch in the parallel for-each > loop, but wholly reproducible when the count is 2 or more. > > Any advice? > > > Try the Hibernate DAO instead of JPA? > > alex > >
