Hi,
I have a cluster scenario like the following:
Two nodes cluster with each node running an instance, and I have a batch script as an generic application resource running on the active node. When there is a public adapter failure on the active node, the script will detect the network failure and kill the instance on the active node and exit, the cluster will then failover the resource(the script)to the standby node. I need to kill the instance on the active node only when the public adapter failure occurs to prevent the split brain from happening.
The script works fine as an generic application alone, but when I make the application as a virtual server(which includes three resources: the virtual ip address, network name, and the batch script), I got the problem as follows: when the public adapter failed on the active node, the resource monitor will detect that and automatically failover the whole resource group to the standby node, the batch script do not have enough time to kill the instance before it has been failoverd to the standby node.
Is there anyway to make the failover happens after the script exits gracefully, or is there anyway to do the post offline action on the failure node(like kill the instance on the failure node after the resource has been failover to the surviving node)?
Thanks in advance !
Regards
Jing
I have a cluster scenario like the following:
Two nodes cluster with each node running an instance, and I have a batch script as an generic application resource running on the active node. When there is a public adapter failure on the active node, the script will detect the network failure and kill the instance on the active node and exit, the cluster will then failover the resource(the script)to the standby node. I need to kill the instance on the active node only when the public adapter failure occurs to prevent the split brain from happening.
The script works fine as an generic application alone, but when I make the application as a virtual server(which includes three resources: the virtual ip address, network name, and the batch script), I got the problem as follows: when the public adapter failed on the active node, the resource monitor will detect that and automatically failover the whole resource group to the standby node, the batch script do not have enough time to kill the instance before it has been failoverd to the standby node.
Is there anyway to make the failover happens after the script exits gracefully, or is there anyway to do the post offline action on the failure node(like kill the instance on the failure node after the resource has been failover to the surviving node)?
Thanks in advance !
Regards
Jing
