After IPL all CICS regions attach to same DB2 subsystem

We run three CICS regions and two DB2 subsystems on one LPAR. Normally they connect through the DB2GROUP and spread load. After last weekend IPL, all three regions piled onto one subsystem and the other sits mostly idle until we bounce CICS.

Anyone seen this? Looking for a way to keep attachment balanced without manual restarts every IPL.

Check the DSNCGRP and which member comes up first after IPL. We had a similar case where the first region to open the group locked the pattern for the day.

Still testing. Opened a case with IBM. Will post back if we get a clean answer.

We stagger CICS colds by region in the automation script post-IPL. Hacky but stable.