Quorum Resource

The quorum resource plays a crucial role in the operation of the cluster. In every cluster, a single resource is designated as the quorum resource. A quorum resource can be any resource with the following attributes:

The quorum resource is used to guarantee that only one set of active, communicating nodes operate as the cluster, acting as a tie-breaker in the unlikely event that cluster nodes are running, but cannot communicate. To maintain a cluster's membership, the Cluster Service relies on the frequent, timely exchange of messages with its peers on other nodes. Although unlikely, it is possible for a failure to occur of all possible networks connecting the cluster nodes. Such a failure could cause the cluster to divide into two or more sets of active nodes that can communicate within the set but not between sets. Each set would assume that nodes in the other sets have failed and attempt to take control of their resources. To prevent this scenario, the node that gains access to the quorum resource is the node that is granted the ability to form the cluster.

The quorum resource maintains access to the most current version of the cluster database and to essential recovery data. When a cluster node fails, changes to the cluster database are written to the quorum resource. To realize the importance of this function, consider the situation where Node2 in a two-node cluster fails. If Node1 goes down before Node2 is brought back online, Node2 must gain control of the quorum resource to be made aware of Node1's changes. Node2 must update its private copy of the cluster database with the changes from the recovery log maintained by the quorum resource. When the quorum resource is out of disk space, all changes to the cluster database are prevented.

Whenever a node attempts to form a cluster, it first tries to gain control of the quorum resource. As other nodes join the cluster, control of the quorum resource can be traded among them, as long as they remain in communication. If communication with the node that controls the quorum resource fails, then the surviving nodes must arbitrate for control. The node that gains control of the quorum resource and the nodes with which the controlling node can communicate will continue to operate as the cluster. Nodes that were previously active and now cannot communicate release all of their resources and halt operation immediately.

The only type of resource defined by Cluster Server that can act as a quorum resource is currently the Physical Disk. However, third-party developers can create their own resource types to be quorum resources as long as they are able to perform arbitration and act as storage devices. It is recommended that clusters use a stripe set of disks.