Problem

A two node CFT installed on 2 host

After a MNRESTAR (without specifying a node, so, for all nodes), even though both LPARs are back available with one each a Copilot, nodes get restarted randomly on one or the other host.

No control on where the node is restarted.

Issue is the same when doing a STOP then a START, nodes are not always reballanced as expected..

Resolution

This is a known issue that has been fixed on all versions supporting the multi-node configuration.

The fix is not in the base versions for V3.1 and V3.2 but in the Service Pack.

It is in the base version for V3.3.

The issue is about the WLM for the sysplex taking control when starting a JOB or a STC to start the node.

It makes the JOB not always executed on the LPAR where it has been submitted/started.

The fix consist to add the system affinity in the JOB card of the MNRMON JCL, the parameter SYSAFF=* (or system=(*) ) ensure the nodes will be started where the JOB is submitted.