Setup ACF2 CPF to send updates to lpars that are sharing databases so that they do not duplicate updates.
The recommendation would be to use CCI's Generic Resources
Here is a sample system setup:
PRD1 and PRD2 share a LID database.
DEV1, DEV2, DEV3 share a LID database.
PRD3 is a standalone LPAR with it's own database.
On PRD1 & PRD2 update the CCIPARM to add a SYSPLEX parm.
The SYSPLEX value for these 2 systems would be the same e.g. SYSPLEX(PRODS).
SYSID(PRD1) SYSPLEX(PRODS)
SYSID(PRD2) SYSPLEX(PRODS)
On DEV1, DEV2, DEV3 update each CCIPARM to add a SYSPLEX parm.
The SYSPLEX value for all 3 systems would be the same e.g. SYSPLEX(DEVS).
SYSID(DEV1) SYSPLEX(DEVS)
SYSID(DEV2) SYSPLEX(DEVS)
SYSID(DEV3) SYSPLEX(DEVS)
PRD3 would not have a sysplex parameter.
SYSID(PRD3)
With this setup, you could put a target list on PRD1 & PRD2 of
DFTPSW(DEVS,PRD3).
With this setup, you Could put a target list on DEV1, DEV2, DEV3 of
DFTPSW(PRODS,PRD3).
With this setup, you could put a target list on PRD3 of
DFTPSW(PRODS,DEVS).
With generic resources, if a connection is down, CCI knows by the SYSPLEX values that lpars with the same SYSPLEX ID are connected and CCI will then try to pass the request to the next node in the sysplex.
This is used to try and guarantee that the request only gets sent once to a group of nodes and it gets sent right away.
Each LPAR will have a Control(CPF) OPTIONS record created and the HOME() value will equate to the SYSID() parameter
in the CCI parms dataset.
In addition, each lpar will have all the other lpars defined in NODEDEF records.
Finally, each SYSPLEX definition will have a nodedef record on each system so that commands can be sent to the SYSPLEX ID or the standard NODE ID.