RE: SQL stored procedures and activation groups via JDBC

Why is it a "bad idea"?
Why do you need to "clean up" these programs? I suppose if they're designed
to be cleaned up by some external action like RCLRSC or RCLACTGRP, I'd
compile them with ACTGRP(*NEW) and be done with it. I'd also say that's a
terrible design (depending on the invoker to clean up your resources).
That said, if you are returning a result set with your stored procedure
(i.e. open cursor), I'm not sure the ACTGRP(*NEW) will work... it's worth a
test for that scenario.

Can you describe what the problem really is? Is something not being
committed or being rolled back when it's not supposed to?

Elvis


Celebrating 11-Years of SQL Performance Excellence on IBM i, i5/OS and
OS/400
www.centerfieldtechnology.com


-----Original Message-----
Subject: SQL stored procedures and activation groups via JDBC

Hi, all:

An IBM i5/OS customer is running DB2/400 SQL stored procedures from a PC
via a Java-based client using JDBC ...

This runs in the QZDASOINIT job -- the top level program is QZDASOINIT,
which is itself an ILE *PGM created with ACTGRP(*CALLER) but for some
reason it runs in the default activation group (*DFTACTGRP) ... :-o

The problem is, they have commitment control, triggers, etc., and the
application programs, triggers and stored procedures etc. are all
writtin in ILE RPG IV and compiled using CRTRPGMOD and CRTPGM and set up
to run as "true" ILE programs with ACTGRP(*CALLER) as they were never
intending to run them in the *DFTACTGRP.

Does anyone know how IBM suggests to support this kind of environment,
when accessing stored procedures via JDBC? Apparently, because
QZDASOINIT is running in the *DFTACTGRP, this is causing these ILE
programs to get activated into the DAG, which is of course a "bad idea"
and there is no way to "clean up" these programs, e.g. via RCLRSC or
RCLACTGRP, and so the jobs in the connection pool just keep growing and
growing (in terms of their storage requirements).

Any suggestions, hints or tips would be appreciated.

Thanks,

Mark S. Waterbury