HACMP 5.3 strange behaviour.

[ogniemi]

I can successfuly start two node cluster with two resource groups - each one has its own home node but both can be started on one at the same time if required (e.g. problem with one node).

After the startup, cluster is STABLE and each of two nodes keeps ONLINE its resource group.

What is strange is that when on STABLE cluster (two nodes UP) I run cluster verification+synchronization resource groups takeovers occures and clfindres shows them in state TEMPORARY_ERROR.

The problem did't occure in previous HACMP releases: 5.2, 5.1, 4.5, 4.4, 4.3 - I run a lot of times verification+synchronisation on running cluster and no takeovers occured.

Any idea?

 

[khalidaaa]

does the cluster.log show any thing?

check 4 event that is failing!