3.11.4 Configuration in which executing node and standby node of Application Server are mutually standby

(1) Features of the system configuration

This is one of the configurations in which one Application Server machine in the standby node is prepared for one Application Server machine in the executing node. However, a different J2EE server is operated as the executing node instead of stopping the Application Server in the standby node. This is called mutual standby.

When you deploy a J2EE server of the same type on each Application Server machine and start a different J2EE server, the respective Application Server instances operate as an active node and a standby node for each other. If an error occurs on any of the machines, the nodes are switched and processes of both J2EE servers are executed on the other Application Server machine.

The prerequisites of this configuration are as follows:

• The CORBA Naming Service must be started in the in-process

• When you use a global transaction, you must store the transaction status file on the shared disk

• Operate using Management Server

Management Server is deployed one by one on the same machine as Application Server and is started on both the hosts. The respective Management Server manages different management domains. The scope of the respective management domains is within the Application Server machine where Management Server is deployed.

The following figure shows an example of the mutual node switching system. In this example, the transaction service is used. The shared disk is required when you use the transaction service.

Figure 3‒52: Example of the configuration of mutual node switching system (for using transaction service)

Note: For other legend items, see 3.2 Description of the system configuration.

Features

• The executing node and the standby node of this configuration are in a 1-to-1 ratio. Management Server is deployed on the respective machines.

• Define two virtual hosts in one management domain (one machine). The system is built by considering these virtual hosts as the host of Application Server in the executing node and the host of Application Server in the standby node.

• One Administration Agent controls starting and stopping of the respective hosts in the management domain. However, the IP address allocated to the respective virtual hosts is different from the IP address that is actually used in the operation. Hence, it is defined in such a way that it is appears to operate on a different host.

• The IP addresses for each Application Server operation are dynamically allocated by the cluster software. For sending the request from Management Server to Administration Agent, an IP address that does not move to the other node after node switching is used.

• When you use a global transaction, a shared disk device is required. A shared disk is required for each virtual host.

• The figure shows a LAN divided into virtual host units, but division is not required.

Node switching flow

If an error occurs on any of the Application Server machines, node switching is executed. The node switching is executed if an error occurs in Management Server, Administration Agent, or the Cluster software. Furthermore, node switching is also executed when a command for node switching is executed explicitly, and when an event occurs where the node switching of the cluster software becomes mandatory.

When the node switching is executed, Administration Agent of the Application Server machine in the standby node starts the processes of the logical server that were stopped until now.

For example, as shown in the figure, if an error occurs in Application Server 1, node switching to Application Server 2 is executed by the cluster software, and processes of the J2EE server 1 and other logical servers are started on Application Server by the Administration Agent of Application Server 2.

To Page Top

(2) Processes to be started on each machine

This subsection describes the required software and processes of the respective machines.

To Page Top