Summary
Although the HA-NFS data service example described in this article is a simple case, the methodology presented is pertinent when creating more complicated data service setup scripts. Determining what changes need to be made to local files on each cluster node and what changes are made to the cluster environment, is key to understanding on which cluster nodes setup scripts need to run. By examining the dependencies between various cluster setup operations, you can develop the sequence of the steps required.
Any well planned setup script should check the current configuration of the cluster before it runs operations that change the configuration. Since you cannot assume that logical device names are always identical, even between similar hardware configuration, you should probe the cluster's physical devices to find out what logical device names they are assigned, rather than specifying hard-coded logical names. Tips for writing scripts that check the cluster status and probe devices were provided along with scripting examples. These can be enhanced to match your specific cluster environment.