Glossary

Cluster

A collection of individual servers (nodes) working to present a single system.

Cluster Membership

A dynamic group of nodes that comprise the current cluster, which are updated when nodes leave or join the cluster.

Cluster Resource

A network service-such as the Web, e-mail, a database, etc.-that is managed by cluster software and is location transparent.

Cluster Volume

A specialized location transparent cluster resource that provides access to NetWare file services.

Distributed Lock

The locking of files in a distribution package prior to its distribution. An attempt is made to lock all files in an archive prior to its extraction and distribution. If all files can be locked, the extraction will execute. If not, the extraction and distribution will fail.

Failback

When a node re-rejoins a cluster, the process of migrating a resource from its current location to its previous location. This can be automatic but we recommend that this process be done manually.

Failover

When a node fails, a user defined process of restarting multiple cluster resources on any of the surviving nodes in a new cluster membership.

Fan-Out Failover

Multiple-node failover support. If a node fails, the process of restarting its resources on the surviving nodes in the cluster. This may happen automatically or not, as determined by the administrator.

Load Script

A script that contains the commands to start a resource or service on a cluster server or to mount a volume on a cluster server. For more information, see the Novell AppNotes, "Configure the Load and Unload Scripts of the Virtual Server".

Location Transparency

The property of a server-side resource, application, or network service that masks its physical location to network clients.

Node

A synonym for a clustered server. The first node to join the cluster is the “master node.”

Quorum

A cluster-wide flag, set once only and never cleared, when (1) cluster membership reaches a threshold number of nodes and (2) a timeout expires, relative to the time the first node joins a new cluster. NOTE: By default, resources do not load until quorum has been set.

RAID

Redundant Array of Independent Disks A category of disk arrays (two or more drives working together) that provides increased performance and various levels of error recovery and fault tolerance. RAID can be implemented in software using standard disk controllers, or it can be designed into the disk controller itself.