Topology changes when using computational partition groups
When you use computational partition groups and a database partition group other than IBMDEFAULTGROUP, dropping a host (for example, when decommissioning a worker) before removing that host from its database partition group causes query errors.
Suppose that database partition group NG_1 contains four workers (nodes 1-4). Before decommissioning the worker on node 3, for example, run the following commands:
$ db2 -s "CALL SYSPROC.BIG_SQL_SERVICE_MODE('ON')" $ db2 "alter database partition group NG_1 drop dbpartitionnum(3)"