A cluster in which one node provides high-availability services and the other node serves only as a backup.
A configuration in which only one node is providing a particular high-availability service. The other node is a backup for that service.
A type of high-availability service. It can be a resource or application. There can be many instances of the application class. For example, network interfaces are an application class.
A pair of nodes with shared disk storage and connected for IRIS FailSafe use.
A file created by an administrator of a cluster that specifies information about the hardware and software configuration of the IRIS FailSafe system. It is installed in /var/ha/ha.conf on both nodes in a cluster. These files must be identical on the two nodes.
A cluster in which both nodes provide high-availability services during normal operation.
A configuration in which both nodes are providing a particular high-availability service during normal operation.
An IP address that is not failed over (in other vendors' high-availability implementations this is sometimes called a node IP address). Each interface used by IRIS FailSafe must have exactly one fixed IP address. Fixed IP addresses are configured in the file /etc/config/netif.options. Usually, the hostname of a node is an IP name for one of the fixed IP addresses on the node. Typically, IP addresses used for primary interfaces and private interfaces are fixed IP addresses.
Messages sent between the nodes that indicate a node is up and running.
Client/server applications that have been configured to run in an IRIS FailSafe cluster. If the IRIS FailSafe system detects a failure on the node on which a high-availability application is running, all high-availability applications and resources are shut down on the failed node and restarted on the other node in the cluster.
An IP address that is failed over by IRIS FailSafe (in other vendors' high-availability implementations this is sometimes called a service IP address). It is configured by IRIS FailSafe and used by high-availability services.
The I/O devices (interfaces and disks) that have been configured as highly-available services on the nodes in a cluster.
Interfaces, disks, and applications that have been configured to fail over to the other node in a cluster under the direction of IRIS FailSafe software. High-availability services can be divided into two categories, high-availability resources and high-availability applications.
A hardware connection on a node to a network. Typical interface names are ec0 and fxp0 for Ethernet networks and xpi0 for FDDI networks.
A network address specified as a name (a string) or in internet address notation (X.X.X.X). The use of internet address notation is required in some cases and strongly recommended in other cases because the name lookup time (see the resolver(4) reference page) can be excessive and cause IRIS FailSafe to assume a failure. You can manually configure an IP address to an interface with this command:
# /usr/etc/ifconfig if_name inet ip_address
if_name is the interface name and ip_address is the IP address (name or internet notation). Configuring an IP address to an interface is normally done by the startup script /etc/init.d/network.
An IP address that is an additional network address for an interface. It is also known as a high-availability IP address. When there is a failure on a node, IP addresses move from their primary interface to their secondary interface (on the other node). IP aliases are listed in /etc/hosts. Configuring an IP alias to an interface is done by IRIS FailSafe, according to specifications in the IRIS FailSafe configuration file.
An IP address in name (string) form. Specifying an IP name for a network address in Internet address notation (X.X.X.X) is done in /etc/hosts.
A hardware-level Ethernet address (a physical address). In most situations, MAC addresses are failed over automatically using the IP re-ARP protocol. For the few situations where the re-ARP protocol isn't supported, IRIS FailSafe provides an alternate mechanism for MAC address failover called re-MACing. See the section “Network Interfaces and IP Addresses” in Chapter 1 for more information.
A network interface that is connected to the private network between nodes.
A network connection between the nodes in a cluster that is not advertised to other networks.
A network connection between the nodes in a cluster and the clients who use the high-availability services provided by the cluster.
A mechanism used to change the physical address of a network interface. The change is not persistent across reboots. See the section “Network Interfaces and IP Addresses” in Chapter 1.
The serial line connection between the nodes in a cluster that provides a mechanism for one node to reboot the other node immediately.