Putting the i in RMI?

I'm working on an RMI service that calculates pricing information for a
point-of-sale client, and having quite a bit of difficulty getting this to
run on our System i. I've tried running a simple main method via RUNJVA, a
direct call from QSH, and a shell script, and the most information I ever
get is a JVA0122 - "Java program completed with exit code 1" message. I've
seen this in the past, but it was generally accompanied by a NoClassDefFound
error. I suspect that that's the problem here, too, but I'm not sure,
because in the past I've always gotten a stack trace outlining what was
missing. This is a much more complex Java app than any I've run on i
before, though, so it's possible I'm missing something that should be
obvious. One thing to note is that the first thing the main method I'm
calling does is a System.out.println("ServiceLoader started..."), and I
haven't yet seen that anywhere.

What do I need to do to get more information on why my calls are failing?
Or can RMI services not even run on i5/OS?

I'm running i5/OS V5R4 with JDK 1.4, 1.5, and 1.6 installed. The Java class
was created in NetBeans on my PC and is expecting JDK 1.6.

My shell script:
----------------------
#!/bin/sh
java -Djava.version=1.6 -verbose -cp
/CalcDir/Rmi.jar:/CalcDir/CalcService.jar:/CalcDir/OAUtil.jar:/CalcDir/log4j-1.2.8.jar:/TaxCalcDir/logging/
com.mycompany.pos.rmi.ServiceLoader -startRegistry
echo "this ran"

...which produces this output when I run it in QSH:
-------------------------------------------------------------------------
Finding native method library: QSYS QJVIO16
Finding native method library: QSYS QJVAUSL
this ran

My RUNJVA command:
-----------------------------------
RUNJVA CLASS(com.mycompany.pos.rmi.ServiceLoader) PARM('startRegistry')
CLASSPATH(&CLASSPATH) OPTIMIZE(*JIT) PROP((java.version
1.6))

This gives the aforementioned JVA0122. Please note that the CLASSPATH
variable here is set to the same value used in the shell script above. I've
tried adding OPTION(*VERBOSE) and OPTION(*DEBUG) to this with no luck, as
well as OUTPUT(*PRINT) and different types of optimization.

Thanks for any input you might have.