In a clustered environment, a service can run on all nodes in the cluster. If this happens, you often need to guarantee continuous access to the data used by that application. Imagine a situation where a Samba server is active on nodel, providing access to shared directories that are on a local volume. After a failure, not only the Samba service needs to fail over to another node in the cluster but also the shared files still must be accessible. To guarantee this, you need some kind of shared storage. Not only is the shared storage needed for access to shared data, offered by the clustered application, but it is also needed for access to the shared configuration files that are used in the cluster. If, for example, this Samba server fails over from nodel to node2, the configuration file /etc/samba/smb.conf on both servers should be the same to ensure the Samba server works the same after it has migrated to another server. You can, of course, make sure these configuration files are copied manually; however, it is more workable to use some kind of shared storage solution.
Was this article helpful?