Testing customized resource recovery

Once you have configured the options and synchronized your cluster successfully, you are ready to test that the new options provide the desired behavior.

Testing the fallover action on resource failure

This is the default behavior. When a resource failure occurs (local_network_down or volume group quorum loss), an rg_move event will be run for the affected resource group. You can test this behavior by inducing a local_network_down (fail all interfaces on that network on a single node) or by inducing the LVM_SA_QUORCLOSE error (power off a disk while writes are occurring such that quorum is lost for that volume group).

Testing the notify action on resource failure

Induce the same failures mentioned previously, selecting notify but no Notify Method. Instead of an rg_move event, a server_down event should run. Check the output in hacmp.out.

Testing the notify method

Configure a resource and resource group and specify the notify option for that resource, with a Notify Method. Induce one of the failures to trigger the server_down event. The server_down event will call the Notify Method and any output from that method will be logged in hacmp.out.