Originally posted by: pat.marshall

We have 2 750's, each split into 2 LPARS. The first LPAR on each is devoted to an application and the second is devoted to the database the application uses. We have it configured like this for kernel tweeking for CPU versus I/O. We have built a virtual ethernet between the two LPARs to leverage the inherent high speed capable there.

We are trying to configure something similar to a site failover in HACMP 6.1 so if LPAR 1 on P750 A fails, both LPARs on A should failover to P750 B. The servers are in the same datacenter and are using shared storage on a V7000. Apparently, site failover will only work with non-shared storage, requiring different disk ID's.

Has anybody configured something like this? If so, can you point me in the right direction?
#PowerHA-(Formerly-known-as-HACMP)-Technical-Forum
#PowerHAforAIX

Originally posted by: j.gann

you want a 4-node cluster with rg dependencies.

Joachim Gann
#PowerHAforAIX
#PowerHA-(Formerly-known-as-HACMP)-Technical-Forum

Originally posted by: pat.marshall

Yes, but not the dependencies currently provided by PowerHA.

The dependency that we are looking for is to have a resource group move to its failover node based on another resource group failing over to its secondary node.

This dependency is not available in PowerHA.

Example:
Nodes: NODE1, NODE2, NODE3, NODE4
Resource groups: RGA, RGB
RGA starts on NODE1 and RGB starts on NODE2.
If RGA fails over to NODE3 because of a failure on NODE1 then RGB fails over to NODE4 and vise versa.
#PowerHAforAIX
#PowerHA-(Formerly-known-as-HACMP)-Technical-Forum