Re: Split-brain among HACMP cluster and Oracle9RAC

I am not an oracle expert, but you mention concurrent access to volume groups.

I assume that means you have configured the resource groups as concurrent, also.

Concurrent resource groups will be offline'd if there is a problem on the machine that the resource group is running on. What Oracle does at this point (Freezing databases and whatnot) is Oracle's problem.

If your problem is that the concurrent resource group was not offline'd, then here is some more information:

As for what HACMP might do, it depends on how the topology configuration is set up, as Hajo mentioned.

One thing that Hajo did not mention, is that even though HACMP may decide that there is a network down event, it won't really care unless you have a service IP label defined on that network that is part of the resource group.

You have to work on configuring HACMP some more. The administration guide has information about selective fallover and post event configuration.

Please remember, that this information is just observations based on your description.

If you have opened problems with Oracle and IBM, they should be able to help you more. Received on Fri Sep 22 2006 - 12:34:50 CDT