Glossary

CLI

Command line interface (sgicm-config-cluster-cmd ).

cluster global lock manager daemon

The clulockd daemon, which stores locks on the shared partition.

cluster membership daemon

The clumembd daemon, which performs network heartbeats and checks the liveliness of other members in the cluster.

cluster quorum daemon

The cluquorumd daemon, which computes new membership, implements quorum, enforces I/O fencing, and reads/writes membership information to the shared partition.

cluster remote NFS mount table daemon

The clurmtabd daemon, which synchronizes NFS mount point entries by polling the /var/lib/nfs/rmtab file.

cluster service manager daemon

The clusvcmgrd daemon, which starts/stops and checks the status of services running in the cluster.

control messages

Messages that SGI Cluster Manager software sends between the members to request operations or distribute information to ensure that services remain highly available.

controlled failback

The service will not be moved to a machine that has newly joined the cluster even if the new machine is the preferred member according to the failover domain. The system administrator must manually relocate the service in order to move it back to the preferred member.

disk tiebreaker

If two members cannot talk to each other, they look at the status on the shared partition disk to decide which member should survive and be part of the cluster membership. If the disk cannot be accessed or membership on the disk does not include a given machine, all SGI Cluster Manager processes on the machine exit.

failback option

A failover domain option that is considered when a member rejoins the cluster.

failover

The process by which one member restarts the highly available applications of a failed member.

failover domain

The list of members in the cluster where a service can be online.

failover option

A failover domain option that is considered when a failure occurs and a new target member for the service must be determined.

failover speed

The time it takes to detect a member failure.

GUI

Graphical user interface (sgicm-config-cluster ).

heartbeat interval

The number of microseconds before a heartbeat is sent to all other members in the cluster.

heartbeat

Messages that SGI Cluster Manager software sends between the members that indicate a machine is up and running.

heartbeat timeout

The number of heartbeats missed before a member is declared as failed.

highly available services

Applications that are monitored by the SGI Cluster Manager software. If one member fails, the other member restarts the highly available applications of the failed member. To clients, the services on the replacement member are indistinguishable from the original services before failure occurred. It appears as if the original member has crashed and rebooted quickly. The clients notice only a brief interruption in the highly available service.

IO10

A full-size PCI expansion board that provides basic system I/O capabilities via the PCI bus.

IX brick

System component that provides the base I/O functionality for the system; it contains the electronics and hardware necessary to boot.

L2

SGI system controller used to monitor and manage the server.

local member

The machine being configured.

local XVM volumes

Logical volumes that are local to one member at a time and are not shared across the cluster. They may change ownership upon failover or moving a service.

lowest-ordered

A higher preference for a service to be started on that member.

member

A machine or system partition that is defined as part of a cluster.

multiport serial adapter cable

A device that provides four DB9 serial ports from a 36-pin connector.

network tiebreaker

Ensures that only the member that can contact the tiebreaker IP address can form a cluster. The tiebreaker is the IP address of a machine or a router that does not participate in the cluster. Usually, it is the IP address of a network router that connects the members to the external world (clients).

ordered failover

A failover domain option that causes the service to start on the first member defined if it is available.

partition

See shared partitions and system partition.

peer member

The other member in the cluster, to which the local system controller is connected.

plug-in

The set of software that allows a service to be highly available without modifying the application itself.

primary partition

One of the two disk partitions without filesystems where SGI Cluster Manager keeps configuration, cluster, and service status information. See also shared partition and shadow partition.

restricted failover

A failover domain option that permits failover only to the members listed.

SATA

Serial ATA disk.

service ID

A number that identifies the service (the ID is automatically determined and is not user-configurable).

shadow partition

One of the two disk partitions without filesystems where SGI Cluster Manager keeps configuration, cluster, and service status. The shadow partition is the backup partition. See also primary partition and shared partitions .

shared partitions

The two disk partitions without filesystems (primary partition and shadow partition) where SGI Cluster Manager keeps configuration, cluster, and service status information.

split-brain scenario

Network partition in which both members attempt to form individual clusters.

symlink

Symbolic link

system partition

A machine that is logically divided into multiple servers. Also referred to as an Altix partition.