As the number of nodes in a cluster increases, its availability increases, but so does the chance of one of them failing at some point. This failure event, whether serious or not, suggests that we must secure a way to isolate the malfunctioning node from the cluster in order for it to fully release its processing tasks to the rest of the cluster. Think of what an erratic node can cause in a shared storage cluster—data corruption would inevitably occur. The word malfunctioning, in this context, means not only what it suggests in the typical usage of the English language (something that is not working properly), but also a node, which also includes the resources started on it, whose state cannot be determined by the cluster for whatever reason.
This is where the term fencing comes into play. By definition, cluster fencing is the process of isolating, or separating, a node from using its resources or starting services, which it should not have access...