Submitting a separate job for each such request, in an attempt to achieve multi-threading, in this case, might actually be worse, in overall elapsed time, due to the extra overhead of job start-up, etc.
You might also want to consider starting one or more "server" jobs that all listen to the same request data queue, and then PgmA just calls QSNDDTAQ to post a request to that data queue, with any needed information such as the "key" or invoice #.
Then, you can "tune" the system by starting up more jobs to run in parallel, all listening to the same data queue. Each job, after processing one request, goes back to the top to issue QRCVBDTAQ to receive the "next" request.
OS/400 and IBM i takes care of distributing requests to however many listeners you choose to start up.
Hope that helps,
Mark S. Waterbury
On Wednesday, April 15, 2026 at 04:48:09 PM EDT, smith5646midrange@xxxxxxxxx <smith5646midrange@xxxxxxxxx> wrote:
If I have PgmA and it repeatedly calls PgmB, I know not setting on LR makes the process faster because it doesn't have to reload PgmB into memory every time.
The argument that we are having is related to PgmA submitting PgmB to enable multi-thread processing. Does not setting on LR help in this case because they are different jobs?
For example,
PgmA
Do until shutdown
Read Invoice Header File for unprocessed invoices
If a new invoice is found
Mark the invoice as In-Process
Submit PgmB to process the invoice
PgmB
Process invoice based on invoice number received
Mark invoice as Processed
** Note: no *INLR logic here
Return
VS
Process invoice based on invoice number received
Mark invoice as Processed
*inlr = *on
Return
midrange.com
