My theory (I'm not sure if I said before) is it uses timestamps to generate MSGIDs, so when more than one message is generated in a second, it shares message ids.

That is not a bad theory. Several years ago, our mainframe at work got an upgrade. One of our programs starting misbehaving badly. Turns out the upgrade sped things up to the point where the transaction ids were the
same... the time used in the ids did not go out far enough to keep the transactions from appearing to happen at the same time.

We expanded them to hundredths or thousandths of seconds to fix it.

I also found some COBOL code that did nothing but count to 10 (or so) when it processed each record. I thought it was an orphan bit from some former, removed process. Finally figured out it was there to slow the program down
to prevent a similar issue as the one above. :)

Mike
---
* SLMR 2.1a * "The Metric System is the tool of the Devil!" - Granpa S


--- GTMail 1.26
* Origin: CCO BBS * 502/875-8938 * capcity2.synchro.net (21:1/175.0)